JP2014510323A - Geographically localized recommendations in computing advice facilities - Google Patents

Abstract

The present disclosure provides geographically localized recommendations to users through a computer-based advice facility. It collects recommendations from Internet sources, the recommendations are judged to have an interest aspect and a geographical location aspect, and the collected recommendations are guided by user preferences and users And determining at least one recommendation to the user based on the comparison process and distributing at least one recommendation to the user's mobile communication device, the user based on the computer Using an application that resides at least partially on the advice facility, at least one of viewing, storing, and sharing recommendations is enabled.

Description

  The present invention relates to a collective knowledge system, and in particular, to providing topic advice using a natural language computer based on machine learning through user interaction.

[References to related applications]
This application is based on US patent application Ser. No. 13 / 155,964 filed Jun. 8, 2011, US provisional application No. 61 / 477,276 filed Apr. 20, 2011, February 2011. US Provisional Application No. 61 / 438,684, filed on the 2nd, and US Provisional Application No. 61 / 430,318, filed on Jan. 6, 2011. All of these are merged as references.

  Searching for topical advice online represents an important use of computer resources such as those provided by the Internet. Computer users use a variety of search tools to search for advice on a specific topic, which requires expertise using a search engine, sifting, interpreting, and comparing It will generate a large amount of search results that take time. People are supposed to use natural language to ask for advice in dialogue, so it is very convenient to have a computer-based advice tool that more closely reproduces how people interact with each other It is. Furthermore, advice on topics will change over time, and a static database of advice will soon be delayed. Thus, there is a need for an improved topic advice search function that uses natural language and is continuously improved in content.

  The present invention provides recommendations to users through a computer-based advice facility. It collects the topic information related to the degree of interest in the topic or the collected topic information including the interest aspect, filters the collected topic information based on the interest aspect, and uses the collected topic information by the advice facility using a computer. Determining the degree of interest and providing the user with recommendations related to the topic information based on the degree of interest.

  These and other systems, methods, objects, features, and advantages of the present invention will become apparent to those skilled in the art from the following detailed description of the embodiments and drawings. All documents mentioned herein are hereby incorporated by reference in their entirety.

  The invention and the following detailed description of certain embodiments thereof will be understood with reference to the following drawings.

Fig. 4 shows a list of topics in the system where the user is determined. An example of a question that the system will ask the user is shown. An example of an image question that the system will ask the user is shown. An example of the type of information that the system presents to the user when making a specific decision is shown. An example of a list for a camera is shown. 2 shows a second example of a list for a camera. 3 shows an embodiment of a user home page. An embodiment of an answer remembered by a user is shown. An embodiment of an answer remembered by a user is shown. Indicates a choice that the user will contribute to expertise. The example of a user's question is shown. Fig. 4 illustrates an embodiment of an answer format. Fig. 4 illustrates an embodiment of an answer format. 2 shows an exemplary list of all decisions in a topic. An embodiment of a processing flow of the present invention is shown. An embodiment of a processing flow of the present invention is shown. 1 illustrates an embodiment of a block diagram of the present invention. Fig. 4 illustrates an embodiment of a submitter / expert interface home page. Fig. 4 illustrates an embodiment of an objective question for a user viewing help when making a decision. Fig. 4 illustrates an embodiment of a decision result indicating a specific recommended decision. Fig. 4 illustrates an embodiment of an interface for a user to set the relationship between attributes and decision results. Fig. 4 illustrates an embodiment showing how a user edits a decision result. Fig. 4 illustrates an embodiment showing a prior revision to content and a change between two previous revisions. FIG. 6 illustrates an embodiment showing questions edited by a user. FIG. An embodiment showing an attribute revision history is shown. Fig. 4 illustrates an embodiment of a workshop interface displaying newly added advice areas. Fig. 4 illustrates an embodiment in which the system asks a user a subjective question for learning user preferences. Fig. 4 illustrates an embodiment showing an activity feed of recent activity from a submitter. Fig. 4 illustrates an embodiment of a result display based on multiple dimensions. 5 shows an embodiment showing a plurality of questions and answer results for a user's indeterminate input. Fig. 4 illustrates an embodiment showing an exemplary question to the user asking for user preferences. Fig. 5 shows a news personality Glenbeck similarity profile determined in an embodiment of the present invention. FIG. 4 shows a personality / mertha steward similarity profile as determined in an embodiment of the present invention. FIG. An embodiment in the case of using a third-party API for supporting learning related to a user will be described. An embodiment in the case of using a third-party API to support learning about the user and aim at a response returned from the user's question to the user is shown. Fig. 4 illustrates an embodiment for determining an unknown user's preference using an Internet social interactive graphic representation. Fig. 4 illustrates an embodiment for improving user preferences and preference profile creation. Fig. 4 illustrates an embodiment of an advice facility using the web to interface with a graph configuration. 3 illustrates an embodiment of an interest recommendation processing block diagram. Fig. 4 illustrates an embodiment of a local discovery applied visual representation of a recommendation to a user. FIG. 4 illustrates an embodiment of a linked detail local discovery applied visual representation for recommendation to a user. FIG. FIG. 6 illustrates an embodiment of a visual discovery representation for recommending to a user. Fig. 3 illustrates an embodiment of a restaurant review. An embodiment of an interest recommendation processing flowchart is shown. FIG. 4 illustrates an embodiment of a geographically localized recommendation process flow diagram.

  While the present invention will be described in the context of particular embodiments, other embodiments will be understood and included by those skilled in the art.

  All documents referred to herein are hereby incorporated by reference.

  The present invention asks the user 1314 a question 1320 and based on the user's answer, the system proposes decisions such as recommendations, diagnoses, conclusions, advice, and the like. Internally, the system uses machine learning to optimize which questions 1320 to make and what decisions 1310 to make at the end of the process. The system learns by feeding back the final decision whether the decision 1310 was useful. Useful solutions are enhanced and associated with questions 1320 and answers 1322 made as the process progresses. When the user 1314 says the decision 1310 was useful, the system remembers which question 1320 was asked and what the answer 1322 to each question 1320 was, and finalized these questions 1320 and answers 1322. Associate with. These associations are the basis for machine learning that learns over time which question 1320 to ask when the user 1314 next comes to the system.

  For example, the user 1314 may seek advice on picking up a bar to go. The system may ask the question “How old are you?” And get the answer “I am in my thirties”. Eventually, the system may show the user 1314 the decision “Kelly's Irish Bar”. Suppose user 1314 says that this decision is useful. The system increases the link between the question "How old are you?", The answer "I am in my thirties", and the decision "Kelly's Irish Bar". The next time the user 1314 comes to the site looking for bar advice, the system is more likely to ask the user 1314 the question "How old are you?" This is because this question 1320 has been useful to the user in the past. If user 1314 responds to question 1320 in the same way as previous user 1314 (says "I'm in my thirties"), the system believes that the final decision is "Kelly's Irish Bar" To strengthen.

  The system builds a profile of each user's preferences, aesthetic preferences, etc., and learns through feedback about which decisions 1310 are likely to be in which type of people. Alternatively, an expert user may specify which type of decision 1310 is likely by what type of people. Learning the user's favorite profile may occur by a separate process from the question 1320 and answer 1322 dialogs asked by the system on a particular topic. For example, the user 1314 may separately tell the system their preference choices for different question and answer dialogs specifically designed to understand the user's aesthetic preferences.

  The user 1314 may not want to spend time teaching the system all of their favorite preferences; instead, in the context of the user 1314 making one specific decision 1310, all preferences An expert may specify which of the questions 1320 is the most important preference question or the system learns. From the territory of all questions, the system may know about discovering about preference profiles, for example, the system is optimal when user 1314 is trying to find a sedan under $ 25,000. You may have learned that there are two specific questions 1310. Alternatively, there may be a completely different set of three favorite questions to ask users 1314 who are interested in SUVs over $ 45,000.

  Users 1314 may also state in the system only about their preferred preferences, not about any objective questions. In this case, the system may provide a ranking of all decisions 1310 in the area of advice based purely on preference. So, rather than saying that the user 1314 wants a $ 200 compact camera, what the user 11314 would be doing effectively is that other urban 35 year old men who prefer computers over sports To say that they want the camera they want. The user 1314 uses the search interface to select an area of advice that is explicitly labeled as “a camera for men in urban thirties”, instead of an area that says “Which camera should I buy”. Selection may indicate this preference. Alternatively, the user 1314 is interested in making a decision 1310 about the camera, chooses not to answer any questions in the Q & A dialog from the system, and the system recommends the camera to the user 1314 It may only have subjective information about the user 1314 used in Alternatively, the user 1314 may answer the question 1320 in the objective and subjective dialog, and the system may recommend the camera based on objective and subjective data about the camera. .

  The user may also enter new questions, answers, and final decisions. The system may try a new question 1320 to see if future questions 1320 make the question 1320 useful to help these users. For example, a bar recommendation service user 1314 may contribute to the question “Do you prefer a loud place or do you prefer a quiet, relaxing situation?”. The system may ask this question 1320 for future use of the bar recommendation service and decide to observe the correlation between the answer to this question 1320 and the recommendation that the user has found useful over the process described above. . On the other hand, the user 1314 may provide a question 1320 that has no value to the user's help. For example, user 1314 may provide the question “Do you have a Canon camera?”. The system may try this question 1320 for a future user and find no correlation between the answer to this question 1320 and the recommendation of the bar that the user deems useful. In this case, the question 1320 may not be asked so much because it does not indicate this or whether the other recommendation is useful.

  The system may continue to ask questions 1320 until it feels highly confident about several possible decisions. The system may also quit early if it already has too many questions 1320 and feels at risk of frustrating the user. The system may also ask at least a minimum number of questions 1320 to avoid that the user 1314 feels that the system is probably not asking enough to make an intelligent decision.

  The system may have a mechanism that allows incorrect answers from users. An incorrect answer may be made because the user 1314 does not understand the question, does not understand the answer 1322, or does not know the answer 1322 to the question. If many of the answers given by user 1314 support a particular decision, the system may make that decision even though not all user answers support that decision.

  In an embodiment, the present invention applies at least one question 1320 and answer 1322 between the system and the user, the decision to the user, and machine learning used to improve the decision. The system is an improved method of generating questions 1320 and answers 1322, an improved method of providing decisions to the user, an improved method of utilizing questions 1320 provided by the system and machine learning to improve the decisions, These functions can be used separately or in combination as a stand-alone system or incorporated into a third party system as an improved function. In embodiments, each of these improvement features may utilize some form of machine learning, as described. For example, the system may provide an improved way to perform a question 1320 and answer 1322 session with user 1314 by learning under what circumstances user 1314 is looking for specific information. I don't know. For example, the system may learn that the climate is a condition that the user has different preferences depending on the time of day and weather conditions. When it is raining day one and user 1314 is looking for a movie, user 1314 may be more likely to be looking for a movie ticket and where the movie is doing. When it is raining all night and the user 1314 is looking for a movie, the user 1314 may be more likely to be looking for a movie description. In other examples, the system may provide a choice vs. single decision 1310 based on age, during the day versus night, and prefer a specific format, based on the user's geographic location, etc. May provide an improved method of providing decisions to the user, such as learning to prefer a large number of questions 1320 prior to presenting the decision 1310. In another example, the system uses information from many professionals based on age and education, uses public opinion when the topic is fashion, user 1314 is younger, and user 1314 is older than Decide what choices to make for the user, such as using more traditional habits, if the topic is personal, or asking more questions 1320 about the user's choices to friends, etc. An improved method of learning to do 1310 is provided.

  In an embodiment, the present invention provides a combination of questions 1320 and answers, provides decisions, and if they are provided by a third party system, which one is provided when not provided by the system Learn to decide. For example, a third-party search engine web application may generate questions 1320 and answers 1322 to improve the ability to provide a sorted list from a user search query and augment the keyword search and sorting algorithm. You may want to take advantage of the facilities. In this case, third-party search engines may not be interested in the ability to generate decisions of the present invention because their services provide a sorted list rather than a limited set of decisions. unknown. However, the present invention allows the search engine provider to improve the sorting results for the user based on the capabilities of the present invention, the ability of the present invention to constantly improve the questions 1320 and answers 1322 to the user. In that respect, it provides an important new capability for search engine providers.

In an embodiment, the subject of the first area of advice is specified in the search interface. For example, where would a user 1314 looking for “a romantic honeymoon in Italy” want to go on vacation? What type of vacation are you looking for? Rather than first asking for user questions, such as, the user 1314 leads to a web page that helps the user decide where to go to the honeymoon. Alternatively, user 1314 can search for a specific location in Italy,
1) If a particular location is good for one of their needs (eg, “This vacation is good for honeymoons and romantic gates, not good for family trips”), user 1314 will help determine
2) Make a suggestion to start a dialog to help the user 1314 find another or potentially good place to go on vacation in Italy,
Lead to a web page. Alternatively, user 1314 is searching for a particular product and enters a dialog to narrow down which of these products are best for them. In both cases # 1 and # 2, the information shown is based on how other users answered the question in the decision dialog and gave positive feedback to the decision. Therefore, many people using the “Where to go on vacation?” Topic answered “Yes” to the question “Do you want a romantic vacation?” And then gave positive feedback to “Italy” If so, the system will display that the romantic destination of the user 1314 who came through the search engine is Italy. Alternatively, the user 1314 who entered the decision “Italy” or the question “Do you want a romantic vacation?” Into the system, the answer “Yes” to the question “Do you want a romantic vacation?” Is associated with Italy, Therefore, the user 1314 who came via the search engine explicitly indicated that Italy is a romantic vacation.

  In an embodiment, the present invention uses the facilities of the present invention to ask questions, make decisions, and make better decisions, but learn better decisions, such as using only input from experts. To provide other combinations of subsets. Rather than providing a specific user with a question 1320 and answer 1322 session, the system can learn from the previous user 1314's interaction with the system to make a decision, or in connection with future decisions. To ask users 1314 questions 1320 and answers 1322, but do not learn anything, such as giving rewards or simply filtering the results instead of giving decisions to users 1314 And asks questions 1320 and answers, uses the features of the present invention to learn how to make a decision 1310 and how to make a better decision, but not to the user via question 1320 and answer 1322 , Provide the expert system with that capability, and so on. In an embodiment, the system provides all the questions 1320, answers 1322 elements of the user 1214 session, provides the user with a decision and learns how to improve the decision.

  In an embodiment, a user 1314 entering a question 1320 may selectively specify the dependency and importance of the question. Dependency may control when questions can be asked. Importance may specify the relative importance between different questions 1320 to weight the user 1314's answer. If any decision 1310 results do not match all the answers 1322 specified by the user 1314, the system will consider a high importance question for a low importance question if the system must take tradeoffs into account. You might try to recommend a decision that matches your question. The system may also prioritize asking high importance questions over low importance questions. For example, a user 1314 entering a new question such as “Where do you want to go on vacation in the United States” asks “Where do you want to go on vacation in the United States” before asking the new question “Where do you want to go on vacation in the United States”? May set up a dependency that requires existing questions such as “Do you want to be answered” to the United States?

  In an embodiment, the present invention may provide the system with a user 1314 interface that allows the user 1314 to interact with the system facilities. A system may include several parts, some of which are websites, supervisors, or collections of widgets. A widget is a collection of code or a process and may render a single piece of content on a website. The website consists of interfaces for end users, staff members, and registered users to obtain decisions, edit decisions, and view reports on system performance. A supervisor may be a container for running a widget so that the user 1314 can perform time-consuming data collection and processing before requesting the content to be rendered.

  For example, a widget may collect videos about decisions from the Internet. The widget may search the video for each decision 1310 on the supervisor and store the videos found in the database. If the user 1314 comes to a website and gets a particular decision, the website may request the video widget to render the video and display any previously found video.

  Multiple supervisor instances may run on multiple computers to scale up widget processing. Each widget may run on its own computer. Similarly, many computers may be embedded in a third party website to provide an interface to the system via a web server, instant messaging, voice gateway, email, program-like API, and the like.

  In an embodiment, an attribute may be a combination of a question 1320 and a specific answer 1322 for that question. For example, when the question 1320 is “How old”, and the answer to the question 1320 is “18 years old or younger”, “20-30 years old”, “more than 30 years old”, the attribute May be "How old? 18 years old or younger". The system may operate by learning the relationship between attributes and decisions. If the system asks a question 1320 and the user 1314 gives an answer 1322, the system may get that attribute and look for which decision is associated with it.

  In an embodiment, the system may understand that some attributes are continuous values and others are discrete values. When using continuous-valued attributes, the system is often accepted to recommend products that are less expensive than the user 1314 asks, but it is rarely accepted to offer products that are more expensive than the user 1314 asks. It may be possible to consider more intelligent trade-offs that make sense of what is not possible.

  In embodiments, the relationship between attributes and decisions may be explicitly given to the system or learned from the user, such as a combination thereof. For example, a price attribute of “How much do you want to use? $ 200 or less” might be explicitly linked to a camera that falls within that price range based on data from experts, commerce sites / APIs, etc. . However, the attribute “How do you use a camera? During a vacation” and the relationship between possible vacation destinations may be fully learned.

  When entering new questions 1320, answers 1322, and results, user 1314 may optionally specify the relationship between attributes and decision results. For example, if the user 1314 enters the question “how much to use” in the “what camera to buy” topic, the user 1314 also asks the system that the answer “$ 200 or less” You might specify that it should be associated with cameras X and Y, not Z. Then, if the future user uses the topic “Which camera to buy” and answers the question “how much do you want to use”? There is a possibility that cameras X and Y are recommended more than camera Z.

  After seeking advice from the system and receiving a decision result, the user 1314 may also be given a reason why the particular decision result was recommended from the system. This description may also allow the user 1314 to change the attribute of the decision if the user 1314 considers that decision to have been incorrectly recommended by the system.

  In general, learned relationships include training from users, professionals, employees, automated data supply from third parties, or combinations thereof.

  In embodiments, there are various ways for the system to recommend a solution and select the next question 1320 to ask the user. Possible machine learning systems may be geometric systems such as nearest neighbors, evolutionary systems such as vector machines, stochastic systems, genetic algorithms, decision trees, neural networks related to decision trees, Bayesian estimation, random forest Supports boosting, logistic regression, faceted navigation, question refinement, question expansion, singular value decomposition, etc. These systems include full gameplay (eg, all attributes given by user 1314 before making a decision), answers to individual question / gameplay subsets, positive feedback only, negative feedback only, Alternatively, learning from these combinations may be based on the periphery. In addition, the system may have previously contacted the user 1314, such as questions that have been previously answered, decisions that the user 1314 has liked or disliked, and areas of advice that the user 1314 has sought before. May take into account the interaction. In addition, the system can use the time of day, the date the user 1314 used the system, the user's IP address, the type of client (eg Firefox, IE, cell phone, SMS, etc.), other such data, etc. The factors provided implicitly by the user 1314 may be taken into account.

  In an embodiment, the present invention is left to the system to guess the user's intention and atmosphere, such as explicitly asking the question 1320 rather than learning implicitly based on the user's behavior. So, we will provide a more powerful machine learning system that goes well beyond the capabilities of collaborative filtering. Also, by selecting a question 1320 to ask the user 1314 based on what has already been answered, the present invention may be able to match the nuances that otherwise would have missed. The present invention explains the decision, such as providing a tightening beyond simple extrapolation from past behavior, such as "Other people who bought X, Y, Z also like product A" It may be possible. Rather, the present invention can tell the user 1314 that “user 1314 wants X, likes Y, believes Z, and should do A”. Furthermore, the present invention allows the user to provide useful new questions 1320 and automatically learns in what context the question 1320 is useful. Another area of difference is that the machine learning technique of the present invention provides for tightness in a very variety of areas of interest for user 1314 that creates difficulties when collaborative filtering is applied to non-commodity / media applications. Can do. For example, collaborative filtering may help user 1314 make decisions 1310 about fairly personal topics such as whether to include tattoos, or unusual questions such as whether certain consumption can be deducted from a user's income tax return. It will not be easy to apply to. The present invention is applicable to such applications. In an embodiment, the present invention can use pre-built expert advice mixed with advice learned from a group of users to give the user a decision.

  In an embodiment, the system may include a wiki web interface for editing all data on the system. This web interface may be used to edit / generate / delete questions, answers, attributes, and solutions. Each solution also has a variety of related information, and when recommending that solution, it is shown on the decision page. For example, if a vacation in Cancun is recommended, the recommendation page may show a video about Cancun. All this ancillary data about the solution may also be editable via the wiki.

  In an embodiment, a wiki may be used to edit data collected by a widget that runs on a supervisor. Thus, the widget can collect data in advance, review it, enable processing for ensuring quality by humans, and change the collected data.

  In an embodiment, the system keeps a history of all changes made by widgets or humans. For example, one use of this history may be used to review work done by contractors hired to ensure content quality. Another use of this history may be to prevent widgets from undoing work done by humans. For example, if a widget collects a particular video and a human deletes this video as inappropriate, the widget may use history so that it does not add this video again at some point in the future. it can. Finally, if the data deteriorates or is deleted by mistake, recovery may be possible depending on the history.

  In an embodiment, when a widget discovers new content, it may queue up work in a human workflow to validate and edit that content.

  In an embodiment, to learn, the system may make random or semi-random decisions from time to time because it may not recommend the system to be useful but may recommend something that proves useful . If the system wants to use what has already been learned, it may not make a random choice as to which question 1320 to ask and which decision 1310 to make. There is a trade-off between using what you already know, using, and learning something potentially new, or research. Use makes the user more satisfied, and research makes the system smarter.

  In an embodiment, when selecting the question 1320 to ask, one way to make this tradeoff is to ask the question 1320 that the system is confident that the user 1314 is valid in making the decision 1310, and then , Employ some random question 1320 to ask. Another way to make a trade-off is to have a fixed budget for each user 1314 exchange, with a fixed set of questions 1320 based on usage and the next set based on surveys. Let's go.

  In embodiments, decisions are also investigated or utilized. If the system wants to learn, it will show a random decision. Rather than showing a purely random decision, the system may also show a decision 1310 that is purely surveyed for the remaining conditions while meeting some conditions specified by the user 1314. For example, instead of employing a random camera to show to the user 1314, the system could employ a random camera that meets the user's price requirements. This will allow the system to perform more effective training because it is less likely to show decisions 1310 that are unlikely to meet the user's needs. At the time of the survey, rather than presenting a random decision 1310, the system may present both a usage-based decision 1310 and a survey-based solution, and may obtain separate feedback from each user. Alternatively, the system could introduce a limited number of randomnesses and adopt a decision “like” what is the best guess for the system. For example, the user 1314 may prefer a particular camera to balance the making of reasonable decisions 1310 and still learning new information from the user, but it is not. You might make a prediction that you could recommend other cameras that are similar, but not the same. In an embodiment, the system may or may not identify the user 1314 when asking the question 11320 and making a decision via survey versus usage.

  In an embodiment, the system may be viewed as examining the user for various recommendations. For example, the system may ask the user ten questions 1320 for a Canon SD1000 camera. This can provide a rich set of data for each camera and allow the system to begin creating a list of what types of users 1314 are likely to like this camera. Given the attributes, the system may generate a ranked list of decisions for each attribute, such as those most likely to be preferred to those most likely to be preferred. For example, the system may generate a list of cameras in the order most likely to be liked by people who responded “How old are you? Above 50”. This is shown by the system as the top 10 camera for users over the age of 50. The various top 10 lists will be created based on system data. These lists are combined to create a new list. For example, given a ranked list of cameras for the attribute "How old are you? Above 50" and another list of attributes "Why do you buy a camera? , A new ranked list of cameras for “users over 50 who want travel cameras” can be generated. Such a combination of the top lists may be generated in advance, or may be generated on demand while asking the user 1314 one by one in order to select a new top list.

  In embodiments, these “top lists” can be used for a variety of purposes. A user may not want to answer a series of questions 1320 before receiving a decision. Rather, they can see these lists and find the goal decision. The system may have a number of top lists, such as one thousand or ten thousand, each with its own web page. In addition, these pages may contain a large amount of content that is indexed by a search engine and directs the user to the system's website. Alternatively, the user 1314 may use the system's own search interface to find areas of advice that it wants to make decisions. Various top lists may be used to shortcut the dialog by implicitly answering some questions 1320 in the dialog based on the top list. For example, there is an area of advice called “Vacations”, and there is a top list called “Romantic Honeymoon Vacation in Italy”, which is a list of some of the already answered “Vacations” dialogs such as: As a shortcut or gateway to the “vacation” topic: “Where do you want to go to Europe?”, “Where do you want to go to Europe? Italy”, “Would you travel on a special occasion? Yes". “What is a special opportunity? Honeymoon”. This is another interface that does not involve the question and answer dialog for a user 1314 seeking advice using a conventional search interface.

  In an embodiment, various pages of a site may have a self-contained display of information called a widget. For example, the decision page shows how other people who liked this question 1320 answered various questions 1320, videos / images about the decision, links to other websites with information about the decision, user 1314 This decision 1310's personal pros and cons based on how you answered the question, a list of other similar decisions, a list of other decisions that would have been made if you answered question 1320 differently, this decision Has a widget that shows a list of rewards / honors for (e.g., recommended consumer reports).

  In an embodiment, the system allows a user to navigate within a decision territory (eg, camera, vacation destination) along a dimension that is not normally available. For example, not only the camera is displayed and the user 1314 not only says “Show me a higher / cheaper camera”, but the system also shows “Show a popular camera to young people”, “Good for travel, You can make the user 1314 say "Show me a non-fashionable camera." Dimensions such as “style”, “good for travel”, “not good for young people” are side effects of learning what is a good decision 1310 when the user is asked questions 1320 and given these answers. Will be generated.

  In an embodiment, navigating along other dimensions, rather than the user 1314 selecting an area for advice and performing a dialog, could be used as a starting point for the user 1314. The user 1314 can begin to interact with the system by using a search interface or an external search engine to search for specific decision results such as product names and travel destinations. The system presents user information about that particular decision result, lets the user 1314 examine other decision results, causes a dialog to specify what the user 1314 is looking for, and about this particular decision result The information learned by the system (by machine learning, expert advice, or some combination) can be shown to the user 1314. For example, the user 1314 can use the search interface to navigate a web page that shows information about the Canon DS1100 camera. The system shows other cameras that people still looking for Canon DS1100 cameras would like, along with the dimension of non-traditional features such as cameras suitable for taking photos at sporting events to users 1314, Can find similar cameras, “good for travel”, “not good for those who are learning photography”, “can buy cheaper than $ 200”, “preferable to those who take pride in favoring good design It can also indicate that the system knows about the Canon DS11100.

  In an embodiment, another possible interface is to show the user a list of decisions and display a brief description of how each decision 1310 was made. For example, when recommending a camera, the system shows three cameras, one is “cheap”, one is “great telephoto”, and the other is “good for travel”. This can help the user 1314 see options that would otherwise have not been seen based on how the question 1320 was answered by the decision 1310.

  In an embodiment, the user will be asked about different types of questions, such as questions 1320 (price, color, etc.) about recommended items, questions 1320 about them. The system will distinguish users along dimensions such as psychological and demographic dimensions. Predictable user characteristics include user age, gender, marital status, country / city life, frequency of going to church, political affiliation, aesthetic taste, sarcasm / humor sense, social Economic background, preferences, cleanliness / things scattered, how planned the future is, etc.

  In an embodiment, it would be difficult to ask the question 1320 directly, so the system would try to measure what is relevant. For example, rather than asking about revenue, the system may ask where the user 1314 prefers to shop (eg, Wal-Mart, Target, Saks, etc.). The aesthetic sense will be determined by showing pictures of art, living room, clothes, etc. and asking if user 1314 prefers that style. In an embodiment, a photo is used for a question (the answer is how you react to this photo) or “which of the following is most similar to the clothes you want to wear. It is used as an answer to the question 1320 such as “Is it?”

  In an embodiment, the system will group the questions 1320 by whether the question is for the recommended item or for the user. What kind of questions 1320 are the system asking to help the user understand the value of the question 1320 being asked, otherwise surprising and potentially aggressive? Would explain. The system also chills the user 1314 by telling the user 1314 how many questions 1320 are left while the question is being asked, what decisions 1310 the system is already making, etc. Will display other messages.

  In an embodiment, an instant messenger (IM) system would provide a natural interface for system question 1320 and answer 1322 dialogs. For example, a user 1314 may invite a system to their “friend list” and initiate a dialog on IM to get a decision 1310. The system eventually instant messages the first question 1320 to the user until the system instant messages the decision link to the user 1314 or directly instant message the decision 1310 name to the user. The answer 1322 may be returned as an instant message. In the embodiment, other forms of communication such as a mobile phone, SMS, and e-mail can be used.

  In an embodiment, a system such as an application form may be embedded in a third party website. For example, the system may be placed on a website that sells cameras and make recommendations to recommend cameras relevant to the user. Alternatively, after the user 1314 searches for cameras and obtains a list of potential cameras that are of interest, the system may ask a question 1320 from within the list of cameras to help the user 1314 make a decision. Absent. For example, if all of the cameras that the user 1314 was thinking about were good for travel, the system would not ask how the user 1314 would like to use the camera and the system would have a replaceable lens. You may understand that asking if you want can be used to recommend one over the other.

  In an embodiment, the system will make decisions in multiple topic areas, such as: For example, merchandise (eg, camera, TV, GPS / navigator, home audio, laptop, bath and beauty, infant, garden / outdoor, car, jewelry, watch, ornaments, shoes, etc.), travel (eg, where Go to, where to stay, where to go, what to do there, etc., economy (eg which property, refinancing, which credit card, what is subject to tax deduction, what type of pension Savings or allocating assets for investment), gifts for various holidays and occasions, other, date-based decisions (like what to wear for Halloween), personalities (eg about user personality) About their relationship, their career, etc.), recommended pets, drinks and other things to spend the night, books, cinemas, movies, sounds , Concerts, TV shows, video games, where to eat, where to order, celebrities related to celebrities most similar to user 1314, gift recommendations, where to live, which TV programs to watch ,Such.

  In an embodiment, the system is a technology / IT (eg, computer, software, printer, home networking, wireless, business network, performance issue, etc.), medical / health, automobile, human relationship or human relationship issue, home And will be used to diagnose problems such as building problems.

  In embodiments, the system user is either unspecified or logged in. A logged-in user 1314 is a person who has created an account on the site. The logged in user 1314 will also have a profile page about them. The contents of the profile page are basic information about the user (nickname, photo, etc.), received and liked decisions, in that topic area the user 1314 did not answer the question 1320, but the system predicted that the user 1320 would prefer A list of facts about the user 1314 given by the user 1314 (e.g., given the user's age or once, so that the user does not need to repeat each time the user 1314 uses the system for a decision) Aesthetic preferences stored over different times, etc.), a list of jobs the user 1314 is allowed to do via the wiki and the system considers interesting (reviewing content submitted by the new user 1314 , Spelling of content submitted by user 1314 To modify the scan, such as reviewing the new content that has been discovered by the widget), it will include such other users who answered the same way in question.

  In embodiments, the user also has various names, ranks and levels that affect what the system can do. For example, a user may be given the name of a “moderator” in that topic that can edit certain aspects of the topic. Ranks and names were given manually, how many decisions were given, how many new questions 1320 or solutions were provided to the system, how many tasks were accomplished using the wiki, and some questions on various topics Given by automatic means, including based on how well the answer 1322 was given to 1320.

  In an embodiment, a non-logged-in user will not benefit from using the system with a large selection of aesthetic or preference-based preferences already registered in the profile. Based on learning and manual training from logged-in user 1314, the system will select an aesthetic question to ask in the question dialog when a non-logged-in user seeks advice in a particular topic area. For example, based on a logged-in user answering questions about his / her preferences and giving feedback on which car he likes / dislikes, the system will allow non-logged-in users to choose between Toyota and Lexus The user will learn if the question 1320 about enjoying a gourmet is useful. By using learned or manually specified attribute relationships by logged-in users, the system will adjust whether to recommend Toyota or Lexus to non-logged-in users.

  In an embodiment, the system learns from the user submitting feedback on the decision. Some users may give false feedback, intentionally or unintentionally. For example, vendors may try to compete against the system to get their products highly recommended. Alternatively, a user 1314 who does not do much about video games may recommend video games that are not really good video games. The system will attempt to filter out feedback from these users in a variety of ways. The system will limit the number of feedback that a particular user 1314 can submit (and have a high limit if the user 1314 is logged in or has a higher rank / name). The system may also narrow feedback based on how well the user 1314 responds to a “test” question 1320 in the Question 1320 & Answer 1322 phase to test the user's knowledge of the subject matter, Weighting or weighting feedback from knowledgeable users more heavily than non-knowledge users. The system may also require the user 1314 to take a “captcha” (a fully automated public Turing test that communicates separately to computers and humans) before feedback is taken into account or a decision is received. unknown. The system will also look at a series of answers given by the user 1314 and weight the user feedback based on the series of answers. For example, if the user 1314 always clicks the first answer 1322 or the user 1314 clicks in a very unlikely manner, the system will weight the user's feedback low. Finally, the system will decide to change the weight of the user feedback based on the previous gameplay history and not show decision 1310. For example, if user 1314 tries to get camera decision 1310 at the tenth time, the system will lighten the feedback weight than at the ninth time.

  In an embodiment, the system will include search engine optimization (SEO), a process that improves the ranking of the system's website within the main search engine. This process understands the competition within the search engine to find out the keywords the user is searching for and searches for these words, causing the page of the site to come out, and how the search engine Some, mostly automated, such as understanding what changes should be made to the system's website to understand how to rank and increase the site's ranking for general search Will be broken down into steps.

  In an embodiment, finding the keyword the user is looking for uses data about historical searches licensed from a third party data provider, looks for other websites to see what words are used, etc. You can find it in different ways, such as using keyword suggestion tools, such as those offered by Google and Yahoo. Once these keywords are found, the system places content on the system's site for these keywords so that they can bet on these words via search engine marketing (SEM) and get future search traffic. You can use this data in many ways, such as creating or seeing how your competitors are using these keywords.

  In an embodiment, the system is doing what other sites are doing, applying keywords to search engines, and who is advertising for each keyword and what is the top natural search result for each keyword. You will understand how it is ranked in search engines. Sites discovered by this process will be searched to find more potential keywords. The system may decide to develop new content or avoid the market based on this competitive information. If there are only a few highly ranked sites in a content area, the system may develop content for that area.

  In an embodiment, the system may understand that paid advertisements that bring users 1314 to a site are relatively cheap in some topic areas of the site and expensive in others. Thus, the system helps these users 1314 by advertising and giving decisions on low-cost traffic, and these users 1314 use the high topic area system to advertise and buy traffic. You might try to recommend. For example, the system will advertise to people who want to find out which dog food is good, help these users 1314 decide which dog food is good, and where to buy pet medicine May lead them to find out. The latter topic area may be expensive due to the expensive advertising rate for the system to draw traffic, and the former topic area is for customers who want advice on what kind of dog to buy Because there are few existing companies that are competing for, it may be relatively cheap.

  In an embodiment, the system examines the relationship between the sites that come up when performing a search and the factors between these sites to determine how the search engine is natural (unsponsored) search results. You will understand what you are ranking. Possible factors that correlate between the higher ranking sites are: site content, number and quality of other sites linked to this site, type of content on these other sites, etc. There will be factors. From the previous step, the system will make a list of site factors ranked, such as the ability to increase the site's ranking in search engines. The system will use this ranked list to make changes to the site to increase the likelihood that the entire site, or specific pages of the site, will be highly ranked in search engines.

  Search engines typically use a keyword index to find documents related to a user's search. In an embodiment, the present invention will also utilize a “decision index” that maps user input to relevant documents. This index can be built automatically, or can be manually built by an expert, or the index can be feedback from different types of users who make implicit or explicit decisions to train the system And so on. Search results that use the decision index will be displayed as a list of documents, a single document, etc.

  Referring to FIG. 1, an embodiment of a list of topics 102 in a system where a user receives a decision is shown: camera, mobile phone, coffee and espresso, drink, favorite celebrity, GPS device, grill, Halloween, laptop, Includes personality, tolling, television, vacations, video games, watches and more. In addition, there may be an indicator of the number of decisions 104 learned from user ratings, such as learning from 43,921 user ratings.

  Referring to FIG. 2, an embodiment of an exemplary question 1320 that the system will ask the user 1314 is provided. In this example, the user 1314 is asking for a decision 1310 regarding the purchase of the camera and the question 1320 is “How much can I use?”. User 1314 selects from options 204 such as cheaper than $ 200, up to $ 300, up to $ 500, higher than $ 500, or not minded. In addition, there will be 202 instructions to ask how many questions 1320 are asked, such as “Get the camera decision that people like you like with less than 10 questions”. In embodiments, users provide their own questions, their own answers, their own decisions, etc., and the system will use this information for current or future decision sessions. In an embodiment, if the user 1314 receives other decisions based on less information available from the user, the user 1314 chooses to skip the question 208 and the system supplements the skipped question 208. Thus, another question 1320 is asked to the user, and question 1320 may be a test question or one that does not affect result determination 1310.

  Referring to FIG. 3, an example embodiment of a photo question 1320 that the system will ask the user 1314 is shown. In this example, the system will ask a question 1320 whose answer 1320 allows the system to better determine the personal characteristics of the user 1314. For example, the illustrated question 1320 asks "Which of these causes are you most concerned about?" If the photo selection shows a specific topic, such as pollution, economy, defense, health, etc. Let's go. This question 1320 may target the current user or be inserted into an experimental question. In an embodiment, the user 1314 is an experimental question 302, as shown in FIG. 3, where the question 1320 has a header such as "Please answer experimental questions submitted by other users last." Will be taught.

  Referring to FIG. 4, an example embodiment of the type of information 402 that the system would show to the user 1314 when making a particular decision 1310 is shown. For example, decision 1310 may be for a camera. Here, information about the camera is provided, such as a description, who uses it, the optimal cost of the camera, and whether it is 404 compared to other cameras. In an embodiment, other decisions 1310 are provided with a relative ranking 408, such as by score or percentage matching. The user 1314 will also be asked for feedback 1312, such as asking if the decision 1310 was a good decision. In addition, the user 1314 may be given the opportunity to find more information about the decision 1310, such as more information about the product 410, an optimal price finder 412, a website for further advice, and the like.

  Referring to FIGS. 5 and 6, the user 1314 may have various top lists related to the topic, as described, such as what was shown for the decision, what was shown for the user's request to see the top list, etc. 502 will be provided.

  In an embodiment, the present invention will provide the user with a home page 700 that includes the user's 1314 ID 702, personal descriptions, decisions made in the past, topics to consider in the future, decisions 1310 to be made today, and the like. FIG. 7 provides an example of a user home page 700, such as what the user 1314 will see when the user logs into the system account. Here, the system's recommended recent decisions, a list of common topics 708 to get decisions, a search interface 710 for finding topics, status updates about the benefits that users 1314 will get from contributing to the system, The activity 704, access to the user profile 712, and the like are displayed.

  FIGS. 8 and 8A show examples of user profile 712 pages showing user information and accounts. User 1314 will manage user information 802 such as the user's email address and password. They also answer questions 1320 about themselves, remember this answer 810, and automatically use it when using system decision topics. The user 1314 will also receive a reward 804 such as a “badge” and will see that it has been received for helping other users, contributing to the system, and so on. Some of these rewards are based on the quality of the user's contribution, the amount of contribution, etc. In addition, users are assigned to a demographic group 808 of people who answered the question 1320 as well for themselves.

  In an embodiment, the user may decide to submit expertise 902 to the system, as in the “Educate system” mode. Figure 9 trains the system on various decisions, rate photos and prose quality submitted by the user, find duplicate items and questions, submit new decision topics, new questions on existing topics Examples of various links / pages that a user 1314 can submit, such as submitting 1320, are shown.

  In an embodiment, the user will be asked a question after the system selects a topic on which to make a decision. FIG. 10 shows an example of how the question 1320 is presented 1000 to the user. As shown, the presentation of the question 1320 to the user 1314 will provide different elements such as a topic heading 1002, a photo or illustration associated with the topic 1004, a question, a collection of answer choices, and the like.

  After answering the question, the user 1314 will be provided with an answer 1322 or a decision 1310 related to the user's original question. FIGS. 11 and 11A show examples of how the decision 1310 is presented 1100 to the user and may include the main decision, information summarizing the decision, other decisions, variations of the decision, and the like. Further, the user 1314 may be given an opportunity to provide feedback 1312 to the system, such as whether the user 1314 agrees with the decision 1310. The user 1314 may also have other suggestive topics 1102 such as based on the current topic, the answers given, the history of answers, the user's profile, the history of the user's questions, topics that other users may find useful. Will be presented.

  FIG. 12 shows an exemplary list of decisions 1200 on topics. In the product topic as shown, the “decision” will be which product to buy. In other topics, decision 1310 may be “Yes. Put him on his neck” or “No. Do not tattoo”. The decisions are ranked and ordered based on relevance to the user, how the user 1314 answered the question, how the user 1314 answered the topic question 1320, and so on. In addition, items are ranked by price, name, etc.

  FIG. 16 shows an example of a submitter / expert interface home page 1600 showing recent system 1602 submissions and other user submissions 1604. In the upper right corner is a question for learning the user's preference 1608.

  FIG. 17 shows an example of a question in a dialog with an objective question 1700 system, for example for a user looking for help in deciding what to name a new puppy. is there.

  FIG. 18 shows a decision result indicating a specific recommendation decision 1800 (in this case, named dog Rusty), reviews of this decision from other users 1802 (ranked by similarity to the user, etc.), An example of a suggestive topic 1808 is shown, such as a yes / no button 1804 (such as for receiving feedback about this decision, or other decision areas that the user prefers). In this example, the second and third best recommendation decisions for the system are listed under # 2 tab 1810, # 3 tab 1812. The system will also investigate by recommending a “wildcard” decision, which is a decision taken in part at random. Suggestive topics 1808 are how the system thinks these topics are relevant to the user and / or how much profit the user using these other areas will generate. Is selected based on how the system thinks it can.

  FIG. 19 shows an example of an interface for the user 1900 to set an association between an attribute and a determination result. In this example, the decision result “Rusty” should be associated with the attribute “This name is a name for a female or male?

  FIG. 20 shows an example of how a user edits the content of the system 2000. In this example, the user can edit the determination result. For example, its name, description, URL for obtaining further information, etc.

  FIG. 21 shows an example of how a user-editable content also has an interface that indicates a change between two previous revisions to view the previous revision 2100 to the content. . The user will also undo changes made by other users if those changes are irrelevant or deemed not useful. In this case, this example shows the difference between the two revisions to the decision result in which the explanation of the result has been changed.

  FIG. 22 shows an example showing a question edited by the user 2200. New answers will be added, the order of existing answers will be changed, the question and answer text itself will be edited, and so on. The question will be selectively “locked” to prevent other users from changing it, such as indicated by the padlock icon 2202.

  FIG. 23 shows an example indicating that the editing of attributes has a revision history like other editable contents 2300. This example shows the difference between two attribute-related revisions between the decision result "Rusty" and the attribute "How many syllables do you want in the name? Less than 2, or 3, or even more".

  FIG. 24 shows an example of a “workshop” screen 2400 that initially displays a newly added advice area. In an embodiment, a user with expertise may make additions without the normal user seeing the ongoing process. Content that is considered problematic, irrelevant or of poor quality will be voted on the system and removed.

  FIG. 25 shows an example of a system that asks a user's preferences / subjective questions 2500 to learn preferences and subjective preferences from the user. After answering these questions, the system will show statistics on how other users answered the same question.

  FIG. 26 shows an example of an activity feed 2600 of recent activity by a submitter across a site showing newly added content and expert training.

  In an embodiment, the present invention provides an improved decision to provide the user 1314 with questions 1320 across a broad category of topics, including merchandise, personal, health, business, politics, education, entertainment, environment, etc. Provide a facility that provides a method. For example, the system would provide a decision on everything from whether the user 1314 broke up with her boyfriend, whether to sculpt a tattoo, whether it can be deducted from income tax in addition to product decisions. In an embodiment, the system will provide a decision on any interest that the user 1314 will have.

  In an embodiment, the present invention will provide a decision system that is flexible, changeable and growable. This is made possible in part by the system using a question 1320 and answer dialog to make a decision and getting feedback from the user 1314 so that the system improves. In an embodiment, this approach would be much more powerful because the system asks every question 1320 and thus gets better information from the user 1314 about their wishes. In addition, the user can expand the system, such as entering his questions 1320 and answers as the system asks, and entering new decisions the system will make. The system will automatically try to see if the newly entered information is useful or helpful, use this new information to determine if it is useful, and will not help the user Will ask / use questions / decisions. In an embodiment, this approach could be used to build a decision expert system based on public wisdom for potentially all topics.

  In an embodiment, the present invention will also provide an improved decision facility for user 1314 by providing decisions by ranking across non-traditional feature dimensions. For example, rather than simply ranking cameras by price or size, the system will rank cameras based on how much they like retired people, how sexy they look, and so on. The system will help users search around across these dimensions. For example, the user cannot simply say “I like this camera but want a cheaper one”, but the system says, “I like this camera, but I want something better for learning photos” and “ I'd like them to do things like saying "I like it but it's better for social scenes."

  In embodiments, the present invention provides a variety of web interface, instant messaging, voice, mobile phone, SMS / instant messaging, third party usage (eg, third party widgets, web services sold to third parties), etc. It will use a different user interface. For example, a voice interface is a good match for a system because it uses only a very limited vocabulary when the system simply has to recognize possible answers to each question. Thus, if the system does not understand the user's response, it will simply move to another question 1320 rather than frustrating the user 1314 by re-listening the answer. In other examples, the present invention is incorporated into a third-party site, such as a search for television on an e-commerce website, and the present invention is a widget that helps the user 1314 narrow the results, or The invention will be used as a widget associated with a real estate website to build an MLS query to find a home that fits the user 1314. In an embodiment, the present invention provides a user interface as a physical interface and as a way to provide questions, answers, decisions, and is significantly improved for obtaining decisions on a very wide variety of topics. Would provide the user 1314 with the same method.

  In an embodiment, the present invention is incorporated into a third party product to improve the third party user interface and user satisfaction. For example, some website services provide predictions from past purchase history. In this case, the present invention could examine the user's mood and intentions, such as by asking explicit questions. In the case of a search engine, the present invention can detect when a user 1314 is making a decision 1310 and will begin asking them to follow a question. In the case of forum sites, mailing lists, newsgroups, etc., the present invention will provide improved access to decisions and will provide decisions made by people similar to the user. For example, the present invention can search all forum posts to find people under the same circumstances as the user, and will provide which decisions the forum community recommended to them.

  In an embodiment, the present invention could extend the user interface of an e-commerce web application. For example, the user 1314 may initiate a product search with a keyword search and ask a question 1320 to narrow the results to the best decision 1310 for the user. The present invention could provide a Q & A interface for picking up merchandise once the user 1314 clicks to enter the category page. For example, after clicking a camera on a website, the user 1314 may see a first question. The present invention can rank merchandise along a dimension based on how the user answered the question. For example, the camera is 1322 where people answered the question “What do you need a camera for?” Answer “Travel”, and thus give positive or negative feedback to a particular camera May rank “travel cameras” from optimal to worst. This would allow e-commerce websites to rank the list of camera keyword search results from best to worst travel camera.

  In an embodiment, the present invention detects when a user 1314 is about to make a decision 1310 and switches to a Q & A interface based on the search results from a keyword search to narrow down the results and re-rank to follow Ask up question 1320 and build a keyword search query or ask question 1320 to refine the search query and after asked the question, user 1314 learns feedback based on which link they clicked It would be possible to provide improved search engine functionality. In addition, the present invention implicitly learns about the user 1314, what time of day the system is used, where in the world, what kind of browser, what weather, etc. Will change the ranking based on the implicit facts.

  In an embodiment, the present invention could provide a way in which information is collected and utilized by a user. For example, Wikipedia is a method by which a user provides information that the end user 1314 has to check the accuracy of the information supplied by himself / herself to some extent. Similarly, the present invention could host web applications that utilize user submitted content. For example, rather than learning what the price of the camera is, the web application will allow the user to enter the price of the camera and allow other users to verify it for themselves. In this way, the range of information submitted could be systematically grown as the user interacts with the system.

  In an embodiment, an e-commerce application may provide goods and / or services related to personal preferences and enjoy the benefits of the present invention. For example, there are currently several movie rental web services where the user 1314 selects a movie to be delivered home by mail. Also, decisions are provided to the user 1314 based on what the user 1314 has selected in the past. However, choosing a movie is a personal interest when renting that cannot be determined by past choices, such as mood, intention, weather, going alone or with someone, their personal relationship, etc. Would include. These types of interests can be examined by the present invention through questions and provide a better match for the user's interests at the time of rental.

  In an embodiment, the local search application will be improved by use of the present invention. For example, if user 1314 wants a decision 1310 about where he wants to have dinner, he may search for “dinner in New York” and find a website that includes suggestions for the query. However, this interface is unsatisfactory if the user does not have a clear idea of what keywords to include. For example, the user 1314 may not know a key option for cooking and may not consider searching for “Ethiopian cooking in New York”. The present invention would have the advantage of being able to find out what question 1320 should be done to narrow down the possibilities. In an embodiment, the present invention could help to build a search query.

  In an embodiment, the present invention will provide an improved method of matching up users and professionals, users and other knowledgeable users, and the like. For example, services may be provided to gather users and professionals on different topics. The user will come to the web interface of the service and enter a Q & A session where the best match is determined. As a result of the query, the system will provide a decision where a profile of an expert or other user 1314 will be provided and the user 1314 will be asked if they agree with the recommended individual. In an embodiment, user 1314 will be provided with a homepage that is kept in communication with previous match-ups and sent to friends, rated professionals, and the like.

  In an embodiment, the present invention will provide a platform for community based question 1320 and answer 1322 applications. For example, a user may post a question 1320 to the system and allow other users to reply. In such a system, user 1314 could receive answers from a single user, multiple users, automated systems, etc. that user 1314 can choose which answer 1322 feels correct. This answer 1322 will be kept secret, posted for others to see, posted as the correct answer, provided to the system, etc. In an embodiment, the system uses questions 1320 and answers to further develop the system, provide more accurate answers by the user, sort the answers provided by the user, and return the answers given to the user. Will be filtered. In addition, the user of the system will provide feedback on answers provided by other users, submit filtering criteria to remove incorrect answers, and so on.

  In embodiments, the present invention may be used through entertainment via the machine learning functions described herein. For example, the user 1314 may think or input ideas such as topics, keywords, categories, questions, feelings, etc., and the system will infer what it is from a series of questions 1320 and answers. For example, the user 1314 may consider an object such as a baseball ball and the system will utilize a machine learning function such as a geometric system to give the user a question 1320. A typical question 1320 may relate to a size such as "Is it bigger than a toaster?" These questions 1320 will be answered by the user via multiple answer selection, writing to a blank, true / false, free response, etc. The system will be able to make a guess and will continue the question 1320 and answer 1322 sequence until this guess is provided to the user. In an embodiment, this process may continue for a fixed number of questions, a random number of questions, a user-specified number of questions, a system-determined number of questions, a system-specified number of questions, and the like. In an embodiment, the system will provide the user 1314 with a user interface, such as via the Internet via a website, stand-alone computing device, mobile computing device, telephone service, voice interface, instant messaging service related, text messaging, etc. Let's go. In an embodiment, the system may be provided to a third party, such as as a widget to another website, as an API to a third party application, and so on. In an embodiment, the present invention will use a non-neural network for entertainment applications, such as playing a game.

  In an embodiment, the present invention will provide a system that assists in the discovery of new drugs, which provides assistance in selecting and combining molecules for the system to make new drugs. For example, the system may ask user 1314 for information related to chemical parameters such as dissolution rate, reactivity, toxicity, etc., and combine these with question 1320 to examine the user's expertise in understanding the structure of the molecule. Let's go. As the question 1320 and answer 1322 sequence progresses, the system will give the user 1314 an insight into which molecular structures are stable and can be synthesized. In an embodiment, the process may include a new potential until a new research route selection is obtained to provide the user until the user 1314 gives an improved sense of which molecular combinations will make a new drug. It will continue until a new drug is identified.

  In an embodiment, the present invention will provide an image discovery application that assists user 1314 in identifying images that meet subjective criteria that are not necessarily explicitly aware of the user. For example, a user 1314 may be involved in the development of a company brochure if there is text in the brochure but must select an image that supports the idea or emotion that the text is trying to convey. The user 1314 would in this case have a subjective idea of what type of photo is needed, but not necessarily able to identify the search by keyword. Instead, the user 1314 will first identify the source of the image, such as from a file, database, website service, from a Google image, or from an advertiser image bank. The user 1314 will then be asked a series of questions and presented with a series of images to select. The answers and / or selections selected by the user 1314 may be used to refine the next choice presented to the user from which further questions 1320 and / or image selections may be provided. In an embodiment, this process will continue until user 1314 finds an image to select as the final image. In addition, the system takes the user's “final selection” and selects another similar group of images to present to the user, at which time the user 1314 chooses to continue the selection refinement process. right.

  In an embodiment, the present invention could be used in a newborn naming application where the user 1314 has a vague sense of what name to use. User 1314 is a system of information that helps to learn user preferences, such as questions 1320 about family, friends, education, inheritance, geographical location, birthplace, hobbies, books read, movies watched, etc. The different types of questions 1320 will be asked first, with the intention of providing The system will continue learning by presenting the name preference question 1320 in multiple ways, by rating the name, selecting from a list of names, or answering the name question 1320. In an embodiment, this process will continue until the user 1314 finds the name to select as the final name. In addition, the system obtains the user's “final selection” and selects from other similarly-named groups for presentation to the user, at which time user 1314 may continue the process of refining the selection. Would choose.

  In an embodiment, the present invention will provide decisions on multiple topics, including but not limited to: For example, video game, laptop, vacation, camera, general personality, drink, mobile phone, TV, grill, clock, coffee machine, tolling, Halloween, GPS device, seasonal celebrity, personal hero, presidential election Kids toys, blogs, camcorders, cars, what Star Wars characters you are, credit cards, hair care, skin care, sex and the city, tattoos, specialty, how much you forgive, City to live, dog type, perfume, New York, neighborhood selection, software, desktop computer, DVD player / recorder, cigarette, charity, Broadway show, speaker, home theater system, MP3 player, computer networking equipment, headphones, Mori card, magazine, book, oppic, book, New York Times bestseller, business / casual clothing, franchise, cookware, toys, toys-education, beauty apparel, espresso machine, should go to Greece, fit parents, raise Whether you need to have a drink, have a drinking problem, have your ADD / ADHD child go to the hospital, vacuum cleaner, clothes washer / dryer, is the right thing to work first, a humidifier, Whether you are a good friend, the risk of becoming diabetic, which foreign language to study, microwave oven, car audio, what customers you should be in, wine, army, what army unit, what kind of army Enjoy various types of art, child seats, strollers, infant travel accessories, natural and organic beauty products, makeup, home use Io receiver / amplifier, copier / fax machine, printer, parting with a lover, which greek god you prefer, which game show you prefer, computer accessories, which superpower you have, university, online degree Choose a program, college major, prevent ID theft, hire a personal trainer, buy a car, rent, or have a laser eye surgery How to deal with this, whether to start an independent business, whether children should go to kindergarten, how to make a family trip to NYC fun, how to make a prescription painkiller, a pre-lifetime will, Miles or cash for the next flight, best way to whiten teeth, allow daughters to make up, hypnosis can cure bad habits, ED Alternatives, sleep introduction, prescription allergies, how much to use for wedding celebrations, should I buy extended warranty, should I get SAT or ACT, personal audio accessories, coffee / espresso is delicious, Video game consoles, jeans, downloadable PC games, snacks, vitamins and supplements, which superhero you are, sunglasses, kitchen accessories, pillows, clean accessories, clean bags and cases, sports goods, which instruments are good Whether to hire decorators, electronic readers, where to belong in shopping malls, power washers, small businesses, phone systems, how much to pay tips, try botox, liposuction, skin cancer Risk of house low Refinance, car service (NYC), local beer, gourmet chocolate, enough to save after retirement, entertainment center / TV stand, cook book, electric shaver, keep sending birthday gifts to nieces , Luggage, computer projector, energy / training stick, razor, gourmet ice cream, online dating, newscast, makeup, tools and brushes, beauty mirror and compact, business book, how long to call after the first date Retirement location, external hard drive, universal remote control, walking shoes, life insurance, how immature you are, how bad you are eating, baby beds, diet and diet books, mobile phones Plans, marriage and engagement fingers , I ’m arguing, my kids are over-playing video games, tax adjustments (individual refunds), home-secured pensions, canceling dating schedules, TV shows for kids, kitchen Countertops, bathroom supplies, insect repellents, cancer specialists, hospitals, national chain restaurants, cereals, whether to make children now, hire a nanny, movies, beef pieces, daily calories, compulsive Whether you have a disability, a home air purifier, an automatic air purifier, an iPhone app, a gay / lesbian vacation, a coworker can be invited to date, or you can leave your child under 10 to a babysitter, Sports / energy drinks, TV shows, office furniture, motorcycles, reward kids with good report cards, lawn mowers, stress Whether you are not too close, religion, good first impressions, whether you spend too much time online, should have a new hairstyle, should educate children at home, should use diaper bags, cloth or disposable diapers, Dog toys, my partner cheating, classic books, old parents should quit driving, forget their previous companions, this is greed or love, pedometer and heart rate monitor, chewing gum, Climatic equipment, gasoline additives are good for cars, Orlando theme park, how big turkey should be bought, popular music-new music, self-tanning, tax and money management, software, baby bottle and infant cup, infants High chair and child auxiliary seat, infant tether, toaster and toaster oven, comfortable seat and bed liner, silver food Equipment set, pet carrier and kennel, cheese, kitchen faucet, casual shoes, dress shoes, beauty electronics, enough money for retirement, selection of mutual funds, steak pieces, own D & D alignment, Acne and breakout treatment, bath faucet, outdoor lighting, landscape lighting, lawn mower, aperitif, cognac, gin, lamb, scotch, tequila, bodoka, whiskey, las vegas show, sunshade, running shoes, US MBA program, Terrace and outdoor furniture, kitchen knife, should you be a real fan, car insurance, personal legal counsel, financial advisor, indoor plant selection, delivery service, deductible, pool heater, sofa, street address, Contact lenses, birthday gifts, my own Career has peaked, e-books, door knobs and key sets, snow removal tools, house greening, children's clothes and swimwear, motorcycle helmets, bicycle helmets, juicers, golf clubs, refrigerators, wine coolers, ranges and ovens, Air conditioner, Christmas gift, farewell, rash treatment, diabetes monitor, smoking cessation, what to do behind the hair, hormones for menopause, hiking backpack, school bag, website / domain acquisition, email service, web hosting, Carpets, power tools, tiles, water heaters, outdoor paint, window handling, fireplace screens, indoor lamps, small business legal consultations, branch recipes, ceiling fans, mattresses, Las Vegas hotels and casinos, salsa, Valentine's Love quiz, how much to use for gifts, anniversary gifts, outdoor jackets, casual jackets, camping tents, sleeping bags, tires, adventurous holidays, music downloads, video downloads, wedding dresses, weddings Theme, Manhattan gym, budget hotel chain, golf course, ski vacation, US spa, ETF fund, designer handbag, should go bankrupt, 401k as home down payment, should be seen by psychiatrist, self-defense , Tableware, dishwasher, political party, New Year's goal, cruise line, family vacation, infant food, infant health care products, shaving head, buying T-shirts, online photo service, graduation rings, summer jobs / Internship, where to go to volunteers, home alarm Stem, diagnose relationship problems, whether she / he is addicted to herself, should be adopted, old parents should continue driving, online bank account, barbecue sauce, frozen pizza, recipe finder, Should I present again, body building supplements, home exercise equipment, how many hours to sleep, whether to consider plastic surgery, arthritis risk, heart disease risk, osteoporosis risk, gambling habits Or choose the best dance to learn, bike, cat food, dog food, hobby advancement, martial arts, poster art, outdoor flowers, which muppet you are? , Children's play, how much you are ethical, how to baptize your child, Miami Hotel, US National Park, motor oil, car video, blouse, coat, dress, glasses frame, socks, visit Clothes, jackets, nightgowns, pants, shirts, skirts, hats, telephone-ground wiring, steak house, which delivery method is best, summer cap recommendations, choice of march madness brackets, milk powder, New York bakery, partial Own jet, how confident I am, digital photo frame, need accountant, my child is ADD / ADHD, document shredder, infant monitor, green house, conference call, etc. .

  In an embodiment, as shown in FIG. 13, the present invention will help a user 1314 make a decision 1310 through the use of a machine learning facility 1302. The process will begin with the initial question 1320 being received by the machine learning facility 1318 from the user 1314. User 1314 will be presented with a dialog 1308 consisting of a question 1320 from machine learning facility 1318 and an answer 1322 provided by user 1314. The machine learning facility 1318 will provide the user with decisions 1310 related to the initial question 1304, such as recommendations, diagnoses, conclusions, advice, etc., based on the dialog 1308. In an embodiment, future questions 1320 and decisions 1310 provided by machine learning facility 1318 may be improved by feedback 1312 provided by user 1314.

  In an embodiment, the initial question 1304 asked by the user 1314 may be an objective question, a subjective question, or the like. Question 1320 will come from a broad category of topics, such as topics related to merchandise, personal information, personal health, economic health, business, politics, education, entertainment, environment, and the like. Question 1320 may be in the form of a question to choose from among multiple, yes / no questions, ratings, image selection, personal questions, etc. Question 1320 may be about user 1314, such as provided by another user, provided by an expert, etc. Question 1320 may be from a current dialog 1308 with user 1314, stored from previous dialog 1308 with user 1314, stored from previous dialog 1308 with other users, or from previous answers. It will be based. Question 1320 may be a pseudo-random question, such as a test question, a survey question that helps to select pseudo-random decision 1310 if pseudo-random decision 1310 may prove to be valid. Question 1320 will include at least one image as part of the question. Question 1320 will be along the psychological dimension. In an embodiment, the question 1320 is not related to the user 1314 directly, but is related to the IP address, the user's location, the weather at the user's location, the domain name, route information, or recently downloaded information. , Information related to recent network access, and information related to recent file access.

  In an embodiment, the dialog 1308 provides the machine learning facility 1318 for a narrowed set of decisions, such as a narrowed set of decisions presented to the user or a single decision 1310 presented to the user. Will continue until a high level of confidence emerges. The decision 1310 provided by the machine learning facility 1318 will be independent of the order of questions in the dialog 1308. Decision 1310 provides another decision 1310 if at least one question 1320 of the dialog is ignored, which is based on machine learning facility 1318 with less information from user 1314. It will be different. Decision 1310 will display a ranking of choices for the decision, such as a ranking decision across non-conventional feature dimensions. Decision 1310 will display at least one image associated with decision 1310. Decision 1310 proves that pseudo-random decision 1310 is useful, such as pseudo-random decision is part of the survey system, survey system improves system effectiveness, and machine learning function 1318 learns from the survey. If so, it would be a pseudo-random decision.

  In an embodiment, the feedback 1312 provided is how the user 1314 responded to the question 1320 in the dialog 1308, and how the user 1314 responded to the decision 1310 provided by the machine learning facility 1318. Will be derived in connection with or from. In an embodiment, feedback 1312 will be sought from user 1314.

  In an embodiment, users 1314 will extend the learning of machine learning facility 1318 by entering new information such as their own topics, questions, answers, decisions, and the like. The machine learning facility 1318 will use the new information to determine if the new information is useful to the user.

  In an embodiment, a user interface may be provided for user interaction with the machine learning facility 1318, such as related to a web interface, instant messaging, voice interface, mobile phone, SMS, etc.

  In an embodiment, the present invention will help the user make a decision 1310 through the use of a machine learning facility 1318. Processing begins with the first question 1304 received by the machine learning facility 1318 from the user 1314, and the first question 1304 covers a broad category of topics such as merchandise, personal, health, business, politics, education, entertainment, environment, etc. It will be related to one. User 1314 will be presented with a dialog 1308 consisting of questions 1320 from machine learning facility 1318 and answers 1322 provided by user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310 such as recommendations, diagnoses, conclusions, advice, etc. related to the initial question based on the dialog 1308. In an embodiment, future questions 1320 and decisions 1310 provided by the machine learning facility 1318 may be improved by feedback provided by the user 1314.

  In an embodiment, as shown in FIG. 14, the present invention will help a user make a decision 1310 through the use of a computing facility 1402. Processing will begin with the first question 1304 received by the computing facility 1418 from the user 1314. User 1314 will be presented with a dialog 1408 consisting of question 1320 from computing facility 1418 and answer 1322 provided by user 1314. The computing facility 1418 will provide a decision 1310 to the user 1314 based on feedback gathered from multiple users 1412. In an embodiment, the computing facility 1418 may improve future questions 1320 and decisions 1310 provided by the computing facility 1418 based on receiving feedback 1412 from the user.

  In an embodiment, the present invention will help the user make a decision 1310 through the use of a machine learning facility 1318. Processing will begin with the first question 1304 received by the machine learning facility 1318 from the user 1314. User 1314 is presented with a dialog 1308 consisting of questions 1320 from machine learning facility 1318 and answers 1322 provided by user 1314, and the number of questions 1320 and answers 1322 provided via dialog 1308 is determined by decision 1310. Will determine the quality. The machine learning facility 1318 will present a decision 1310 such as a recommendation, diagnosis, conclusion, advice, etc., associated with the initial question 1304 to the user based on the dialog 1308. In an embodiment, future questions 1320 and decisions 1310 provided by the machine learning facility 1318 may be improved via feedback 1312 provided by the user. In an embodiment, the higher the number of questions 1320 and answers 1322, such as more than 10 questions, more than 15 questions, and more than 10 questions, the quality will increase. In an embodiment, if the number of questions 1320 and answers 1322 is small, such as fewer than 10 questions, fewer than 5 questions, fewer than 3 questions, 1 question, etc., the quality will be good. Let's go.

  In an embodiment, as shown in FIG. 15, the present invention will make a decision 1310 through the use of a machine learning facility 1318. The system includes a machine learning facility 1318 that receives an initial question 1304 from a user 1314, a dialog facility 150 provided in the machine learning facility 1318 that provides a question 1320 to the user 1314 and receives an answer 1322 from the user, etc. Learning facility 1318 provides decision 1310 from decision facility 1504 to user 1314. In an embodiment, the decision 1310 provided to the user 1314 is based on an exchange of dialog 1308 between the user 1314 and the machine learning facility 1318 and may relate to the initial question 1304. Further, the machine learning facility 1318 will receive feedback 1312 from the user 1314 via the feedback facility 1508 to improve future questions 1320 and decisions 1310 provided by the machine learning facility 1318.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with an initial question 1304 that the machine learning facility 1318 receives from the user 1314 via a third party such as a search application, social network application, service provider, comparative shopping engine, media company web environment, and the like. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, future questions 1320 and decisions 1310 provided by the machine learning facility 1318 may be improved via feedback 1312 provided by the user 1314.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing begins with the first question 1304 that the machine learning facility 1318 receives from the user 1314 via a third party search application, and the user 1314 begins with a keyword search in the third party search application and from the machine learning facility 1318. Will be presented with a dialog 1308 consisting of a question 1320 and an answer 1322 provided by the user 1314. The machine learning facility 1318 provides a decision 1310 to the user 1314 related to the initial question 1304 based on the dialog 1308, and the decision 1310 may be returned to the third party search application, such as in the form of a sorted list.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 is presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314, and the machine learning facility 1318 will utilize third party information, functions, utilities, etc. . The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, third party information, functions, utilities, etc. will include an application program interface (API) that allows collection of cost information, product information, personal information, topic information, and the like.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing begins with an initial question 1304 that the machine learning facility 1318 receives from the user 1314 via a third party search application, and the user 1314 begins with a keyword search using the third party search application, and then the machine learning facility 1318. A dialog 1308 consisting of questions 1320 from and answers 1322 provided by the user 1314 will be presented. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, the decision 1310 will be returned to the third party search application based at least in part on collaborative filtering.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnosis, conclusions, advice, etc., associated with the initial question 1304, along with the at least one image, based on the dialog 1308. In embodiments, the images may be photographs, paintings, video images, advertisements, and so forth.

  In an embodiment, the present invention will help user 1314 make a decision 1310 through the use of machine learning function 1318. Processing will begin with the first question 1304 that the machine learning function 1318 receives from the user 1314. The user 1314 is presented with a dialog 1308 consisting of a question 1320 from the machine learning function 1318 and an answer 1322 provided by the user 1314, where the question 1320 is at least partially from other users of the machine learning facility 1318. Will be determined by learning. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnosis, conclusions, advice, etc., associated with the initial question 1304, along with the at least one image, based on the dialog 1308. In embodiments, decision 1310 may be based at least in part on learning from decision 1310 provided by other users of machine learning facility 1318.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, decision 1310 may be based at least in part on collaborative filtering.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, decision 1310 may be based at least in part on collaborative filtering where context is provided via dialog 1308, such as at least one question that provides collaborative filtering context.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. related to the initial question 1304 based on the dialog 1308. In an embodiment, the decision 1310 is based on information gathered only via the plurality of users 1314 of the machine learning facility 1318 and is related to the initial question 1304, wherein at least one of the plurality of users 1314 of the machine learning facility 1318 is: User 1314 associated with dialog 1308.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with a decision 1310 based on the limited initial machine learning facility 1318 knowledge regarding the subject of the first question 1304 that is related to the first question 1304 based on the dialog 1308. Let's go. In an embodiment, the limited initial knowledge of the machine learning facility 1318 is seed knowledge, limited to basic knowledge about the subject of the first question 1304, and limited to basic knowledge about the subject of the first question 1304. Basic knowledge would be expert knowledge.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 provides the user 1314 with recommendations 1310, recommendations, diagnoses, conclusions, advice, etc. regarding the subject of the first question 1304 related to the first question 1304 based on the dialog 1308, where the decision 1310 Based on learning from a combination of home and user input.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 will provide the user 1314 with a category-based decision 1310 on the subject of the first question 1304 related to the first question 1304, such as recommendations, diagnoses, conclusions, advice, etc., based on the dialog 1308. Let's go.

  In an embodiment, the present invention will assist user 1314 in making a decision 1310 through the use of machine learning facility 1318. Processing will begin with the first question 1304 that the machine learning facility 1318 receives from the user 1314. The user 1314 will be presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The machine learning facility 1318 provides a decision 1310 to the user 1314, which utilizes the responses from the plurality of users 1314 of the machine learning facility 1318 to categorize and at least psychological and demographics. It will provide a decision 1310 along one of the academic dimensions.

  In an embodiment, the present invention will provide a response to the user 1314 through the use of the machine learning facility 1318. The user 1314 is presented with a dialog 1308 consisting of a question 1320 from the machine learning facility 1318 and an answer 1322 provided by the user 1314. The question 1320 from the machine learning facility 1318 is applied to entertainment and drug discovery applications. Would be related to the response, such as application to the name of the newborn. The machine learning facility 1318 will provide the user 1314 with recommendations, diagnoses, conclusions, advice, etc. responses related to the subject of the first question 1304 that are related to the first question 1304 based on the dialog 1308. In an embodiment, future questions 1320 and decisions 1310 provided by machine learning facility 1318 may be improved via feedback 1312 provided by user 1314.

  In embodiments, the present invention will provide results based on multiple dimensions, such as results based on text matching from user input, results based on profiles of user preferences, and the like. FIG. 27 shows an example of a search interface showing search results for the user query “fios”. In this example, the ranking is the second ranking according to the user query, which matches well with text matching, the first found decision and decision result (recommendation), and the knowledge the system has about the profile of user preferences Based on the decision result. In this example, “fios” is a good text for the decision result “Verizon BIOS” as a recommendation for “which ISP to use” 2702 and “which US satellite / cable service provider to use” 2704. It is a match and both are the results of this user's # 1 ranking based on the user's favorite preferences.

  In an embodiment, the present invention will respond to the user's indeterminate input and provide multiple questions and an answer “result”. For example, FIG. 28 illustrates a case where the user query is “sub”. The first result is on the topic “Which new car should I buy?”, But with the question “What type of car do you want?” Already answered “sub”. This would bridge the irregular search and the regular Q & A data stored by the system. Further, as shown in this example, the user's top 3 results are displayed for personal use based on their favorite profile. Effectively, the user can ask other questions as shown in the examples, “Which BMW should I buy?” 2804, “What rate model should I buy?” 2808, “Which car or truck to buy? Like “Would you use a tire?” 2801, through a conventional Q & A interface, keyword search was performed and the results were obtained without explicitly answering the question.

  In an embodiment, the present invention expresses a preference to the user if they give answers to the question whether they contradict each other or are mutually exclusive, each individually having a significant impact on the ranking of the user's results. You might ask. For example, a user may have started the “Which New Car to Buy” topic and would like a more practical SUV that is cheaper than $ 18,000, less luxurious, and more practical. The system may want to get the user's preferences as to whether the car is an SUV or cheaper than $ 18,000, which is more important. FIG. 29 is an example of a question to the user asking for preference.

  In an embodiment, the system will learn a set of importance of questions, such as for each user, for each decision result, for each decision result for each answer to the question. For example, the system will learn that when user A buys a small camera, weight is more important than price, but when buying a car, price is more important than anything.

  In an embodiment, the user will evolve the system by adding new decision results. Such a new decision result will optionally include a link to a web page to read more about the decision result. The system will automatically convert and associate with these links for the system to be delegated from the site that the links point to. Further, based on the link submitted by the user, the system recognizes what kind of link this is, and Amazon. If it is something like that from the You will understand how to interpret the code.

  In an embodiment, users will be clustered into groups using dimensionality reduction techniques such as singular value decomposition (SVD), eigenvalue vectors, and other similar methods. The system will display information about why a group of users are clustered together. One way to do this is to find the top X dimension in a low dimensional space where the clusters are most different overall from the general average. The spread of the response cluster distribution from the general mass distribution of each dimension in the subspace is used to rank the dimension in terms of how well it explains what is unique in each cluster. Will be.

  In an embodiment, dimensions in a subspace can be described and interpreted by a number of different features, such as how people answered the question or which decision result they liked. It will not be easy. One way to explain what each dimension means would be to find the questions and answers that are most correlated with the extreme values of the different dimensions and label the dimensions with these questions / answers.

  In an embodiment, one method for clustering users employs the user's initial random grouping and recursively minimizes how different users are within each other in their own cluster. Would move between clusters. After processing recursively several times, the process will be stopped, or the process will continue until an error threshold amount is reached.

  In an embodiment, the present invention will facilitate matching by user similarity. Given a user name, email address, user ID, etc., in some sense, a list of other users that are similar or different is provided. For example, given a Facebook username, it will provide a ranked list of other Facebook users with similar preferences in a specific sense, such as overall or electronic devices. Let's go. In addition, this list may be displayed on Facebook with the similarity to you so that you can ask the question to the person who is once away from the first user in the social graph. Will be limited arbitrarily). In embodiments, a user's similarity may be calculated by asking the user questions about himself, looking at the social graph, and using context such as location, IP address, time, etc. Social graphs are based on a heuristic approach to usernames, email addresses, surnames, names, birthdays, addresses, genders, and other similar information to make social graph users known users of other data sets. Will be used by mapping. Adjacent people may be looked up in the social graph to take into account more people, even if they are farther away from those who are trying to “triangulate”. For example, in order to find a user on Facebook that is most similar to you, to combine a social graph from Facebook with a user who wrote a review on Amazon, and to get a recommendation for a notebook PC in Amazon, You will see your favorite notebook PC. As another example, reviews may be filtered based on people who are similar to them on sites such as Yelp, Tripadvisor, Amazon. This information will be used on Twitter to help users to recommend users to “friends” on Facebook, for example to “follow”. 30 and 31 show examples of similarity profiles that may be provided by the present invention.

  In an embodiment, the present invention will provide an indirect level between recommendations rather than recommending people who like / dislike alternates. The invention is based on what similar users buy based on what they bought on Amazon, where to eat based on similar users in Yelp, Zagat, Foursquare, etc., and what similar people clicked on Would recommend what to click in the Google search results. For example, consider the problem of recommending which new users on Twitter should follow. The present invention looks for all users on Twitter to follow them, knows about each user based on the questions answered about himself, and matches some of those followers with the data set of the present invention. right. This will also match followers with other users, such as Amazon users and Yelp users, to learn other things about them. Based on this, the present invention will make inferences about the followers of a particular user, such as a Twitter user. Here, the new user is asked about themselves and this new user will find which of the existing user's followers is most similar. A recommendation will be generated for this new Twitter user to follow an existing Twitter user whose followers are most similar to the new user.

  In an embodiment, the present invention provides recommendations that immediately reflect new information from users, social graphs, new facts about them, etc., to rerank the recommendations or improve the results, It will facilitate real-time personalization, such as immediately using contexts such as changing locations.

  In an embodiment, the present invention uses the location, time, etc., implicitly to indicate a nearby location where the user eats, drinks, wants to see, etc., and what to do. Matching will be based on the weather, social graph, and other contexts. For example, this process may be used in mobile service applications where location data can be accessed via GPS. Or, the recommendation is that the current weather (eg, if it's raining, people don't recommend the destination only on the terrace), the current time (eg, do not recommend night clubs at 10am) ), Calendar (e.g. knowing when the user is busy or where they must go in the future), social graphs (e.g. recommending places where the user's friends are otherwise exactly equivalent), etc. Will be notified by the part.

  In an embodiment, the present invention will provide natural language questions and answers, such as allowing indefinite or fixed input from the user for decisions or recommendations that they want to help. Input uses automated techniques such as natural language processing to ask the user, ask other users to review the question (ie, "Is this question about electronics, travel, car or other topics?") It will be categorized by.

  In an embodiment, the present invention will find users similar to the requesting user who have expertise in the question category. For example, a user wants advice on a hotel in Los Angeles from people who have similar preferences about the hotel and who are knowledgeable about the hotel in Los Angeles (introductory or public knowledge based on their behavior). It may be. For example, the present invention may alert those similar users to new questions and request assistance to resolve them. You will consider how many questions / notes you have sent before, how many you responded, how useful the reply was. Similar users will be allowed to interact with the requesting user to help determine the requesting user or notify the recommendation issue. The resulting dialog will be stored for use by others and will prompt similar users to index the dialog to be boilerplate to help other requesting users later remember. .

  The present invention will provide a third-party site that has the ability to learn about users, find similar users, and make recommendations that are independent of direct interaction with the user's system. In an embodiment, the present invention is preferred if a third party, such as hosting its own website, prefers if the user does not dialog or interacts directly with the computing facility of the present invention. Will provide preference and preference APIs that will be used to learn about preferences, likes, dislikes, and other attributes of the user. For example, the user is Amazon. go to websites such as com and ask questions about the product. In this example, Amazon will better determine the user's preferences, preferences, etc., so that the third party can better answer the user's questions and provide meaningful responses targeted to the user. You will have an API of the present invention that allows you to create or improve preferences and preference profiles. In an embodiment, the third party, in this example, from a previous interaction on a third party site such as Amazon, or from a previous interaction with another website hosting a similar API, or In order to determine user preferences, preferences, etc. without interacting with the user, such as determining their preferences and preferences from previous interactions with the user, such as direct interaction with the facility of the invention. Will use the API. In an embodiment, the API uses the user interaction to identify other similar users and assists in selecting recommendations for the current user, such other user selection, determination, selection, It will be used by multiple users, such as using recommendations. In embodiments, these other similar users may be associated with APIs hosted by a third party, from other hosted API sites, or may come from facilities of the present invention. In an embodiment, the use of the API of the present invention hosted by a third party site provides significant advantages to the third party site for recommendations related to user preferences, preferences, likes, dislikes, attributes, etc.

  In embodiments, user preferences and preferences may include other similar users, other users connected to users of social networks, other users related through personal or professional activities, friends, family, etc. It will be determined and augmented by other users, such as other users. In an embodiment, this would be done without having to ask the user a question. For example, a user has an existing preference and preference profile, as established by the present invention, which collects information about other users, such as social networks, families, workplaces, or infers It will be improved by doing. In an embodiment, the user's profile is improved by inferring additional preferences and preferences from other similar users or users who have been shown to have some connection with the user, such as via social networks. right. In an embodiment, the user's profile will be improved by selections made by other similar users in product selection, recommendation, etc. In embodiments, the system infers preference profiles by their rating, what they like and dislike, or by analyzing how users tag their user profiles. By using natural language processing, you will learn about known user preference profiles.

  Referring to FIG. 32, in an embodiment, the present invention is a computer-readable medium that, when executed on one or more computers, helps a third-party website learn about a user by using a computer facility 3202. Would provide a computer program product recorded on. This provides (1) a user preference learning API to third party websites 3220 to determine user 3218 preferences when applied to a third party market, which is implemented as an extension of computer facility 3204. , (2) receive third party information related to the third party 3208 market, (3) collect user 3218 preferences, store as user preference profile 3210, and (4) relate to third party 3212 market. The third party website 3220 receives a query from the user and (5) recommends the third party based on the user preference profile and the third party information to help the third party answer the received query 3214. It is carried out the step of providing. In an embodiment, determining the preference will be done using natural language processing. The computing facility will be a machine learning facility. Third-party information consists of at least one of product information from the product manufacturer, product information from the web store owner, price information from other websites, inventory status from the store owner, reviews, comments, and ratings. right. The preference learning API will enable collection of at least one of cost information, product information, personal information, and topic information. In addition, user profile preferences may also be based on information inferred from the user's social network where the user does not receive additional dialogs between the user and the computer facility.

  Referring to FIG. 33, the present invention shows reviews from similar users that target ads, recommend products, show similar people on social networks, and based on the results of similar users clicking the most. Will provide the use of preferences and preference APIs to target responses returned to the user, such as ranking search results. In an embodiment, the present invention provides a computer program product recorded on a computer readable medium that, when executed on one or more computers, uses a computer facility 3202 to help narrow the target to a response to the user. Would provide. This provides (1) a user preference learning API to a third party website 3220 to determine user preferences related to the third party market, which is implemented as an extension of the computer facility 3204; Receive third party information related to the third party 3208 market, (3) collect user 3218 preferences, store as user preference profile 3210, and (4) query from user 3318 on third party website 3212. (5) In the user preference learning API, a response related to the user query is returned using the third party information related to the third party market and the preference of the user 3318 stored in the user preference profile. Step Carried out. In an embodiment, the computing facility will be a machine learning facility.

  The response provides an advertisement to the user, and the advertisement will be based on the user preferences stored in the user preference profile. Advertisements are supplied by a computer facility via a third party, enabled by preferences supplied from the computer facility to the third party, supplied to other users of the user's social network, and so on. The response will provide recommendations for goods, services, etc. related to the third party market. Collecting at least a second user's preferences will generate a user preference profile for the second user and will determine that the second user is similar to the user based on the preference profile comparison. The response will provide a recommendation made by the second user. Collecting preferences for the second user is done from a social construct based on the internet, and a response is provided as information to the user indicating the second user as a similar person on the construct, A social construct based on the Internet would be a social network. Collecting preferences for at least a second user includes a search result selection, the query is a search request, and the response is ranked according to at least a second user search result selection. Will be a search result. Collection will be done by recommendations, purchases and search result selections made by the user. The collection will be from an information source that reveals the movement of the user's location. The information source may be user location information from web services foursquare, yelp, Google, Gowalla, Facebook, etc. The information source will be user location information from the service provider. Third-party information includes product information from the product manufacturer, product information from the website owner, price information from other websites, inventory status from other websites, price information from shop owners, inventory status from shop owners , Reviews, comments, and ratings. The definitive preference will be due to the use of natural language processing. The API will enable collection of cost information, product information, personal information, topic information, and the like. The collection is from user interactions, such as those represented by social interactive constructs based on the Internet, which would be social networks. The response would be a list of reviews sorted by the review author's similarity to the user reading the review. Collecting user preferences may be via a third party website. The collection will be by visiting a third-party website.

  In an embodiment, the present invention allows third parties to target advertisements to users based on user preferences when the user does not interact with the dialog or interacts directly with the computing facility of the present invention. You will use the preference and preference API that you use. For example, the user has previously experienced a system and question and answer dialog, through which the system will generate preferences and preference profiles for the user. In embodiments, the dialog may be provided directly to the facility of the present invention or by a third party API provided by the present invention. Alternatively, the user's preferences and preference profiles may be generated and updated through user interactions, responses, recommendations, reviews, etc., but may not have interacted with the facility of the present invention. . In an embodiment, the system uses natural language processing, such as inferring preference profiles, through the rating of likes and dislikes, or by analyzing how users tag user profiles. Through use, you will learn about known user preference profiles. User preferences and preference profiles may be used to target ads to users, such as ads that match user preferences and preferences. For example, third-party preferences and preference APIs are L. Related to outdoor store websites such as Bean, REI, EMS, etc., the outdoor store will be trying to improve the target of advertising to customers. Customers will visit outdoor store websites and query for items such as hiking boots. The preference and preference API will allow browsing of user preferences and preferences to match advertisement placement to the user's browser. In this example, the user preference and preference profile will indicate whether the user enjoys a trip to New England, likes camping, or has a home with children. As a result, the advertising facility associated with the website will select advertisements utilizing information from user queries, in this example, hiking boots, and information from their preferences and preference profiles. In this case, the ad would be for staying at a White Mountain lodge that combines a user's query for hiking boots and travel preferences to New England. Furthermore, accommodation will be family accommodation due to the fact that the user prefers to travel with the family and attributes that reflect the user's preferences. In an embodiment, third party preferences and preference APIs can improve the targeting of ads to users to allow third parties to increase revenue from advertisers who have made certain advertisements. Will do. In an embodiment, user preferences and preference profiles will be built in real time, focusing on the previous query. Preference and preference will be collected to better target ads to users, such as during purchase, during refinement after product search.

  In an embodiment, the advertisement will be targeted to a user or a group of individuals related to the user based on preferences and preferences inferred via the user's social network. For example, preference and preference APIs used by third parties may be used to establish preferences and preferences for groups, node clusters, etc., as in a user's social network. In embodiments, preferences and preferences inferred from social networks will utilize previously formed preferences and preference profiles, such as by third party sites or by facilities directly related to the present invention. . These preferences and preferences will be used to better target ads to users and members of their social networks. In the example, the third party would like to target the advertisement to the user if the user has built preferences and preference profiles stored in the facility of the present invention. The third party will use information from the user's profile to target the advertisement. Alternatively, the third party may additionally use information inferred from the social network in which the user is participating, such as information related to the topic of the social network, the user's common interest regarding the user in the social network, etc. . For example, a user has a preference and preference profile associated with social network users who are middle-aged, politically conservative, or rural and have hunting as their primary interest. Will be. In this example, the third party will target the ad to hunting equipment, hunting trips, etc., and the ad will be selected based on the user's existing preferences and preference profiles, as well as inferred from the user's social network. It will be. In an embodiment, this may be done without the user having to go through the described dialog, but on a user interaction such as on a third party website, on other websites where the present invention has a third party API, etc. Will be done indirectly by information provided by third parties. In an embodiment, the third party will also use the preferences and preferences from the user to target the advertisement to other members of the user's social network.

  In the embodiment, advertisements to users are targeted based on product selection, recommendation, and the like of users who have similar preferences and preferences. For example, the first user has similar preferences and preferences as the second user, and the first user has existing preferences and preference profiles, and has made certain product selections, recommendations, etc. right. The advertisement will be targeted to the second user based on the first user's decision. For example, a first user may have a profile that indicates that he is aged, retired, lives in California, enjoys a trip, etc. and has previously selected a product for his travel bag. The second user will receive the same travel bag recommendation based on the preferences of the two users and the similarity of the preferences. In an embodiment, this would be done without the need for either user to dialog.

  In an embodiment, the present invention provides a computer program product recorded on a computer readable medium that, when executed on one or more computers, uses a computer facility to help users target advertising. right. This provides (1) a user preference learning API to third party websites to determine user 3218 preferences when applied to a third party market, executed as an extension of computer facility 3204; (2) Receive third party information related to the third party market, (3) collect user preferences and store as user preference profiles, (4) third party websites related to the third party market , Receiving a query from the user, and (5) providing an advertisement to a third party based on the user's determined preference. In an embodiment, determining the preference will be done using natural language processing. The advertisement will be provided by a computer facility. Advertisements will be provided through third parties and will be possible based on preferences provided by computer facilities to third parties. The API will enable collection of at least one of cost information, product information, personal information, and topic information. The decision will be based on information inferred from the user's social network. The advertisement will be distributed to other users associated with the user, such as via a social network.

  In an embodiment, the present invention allows third parties to provide users with similar user reviews when a user similar to the user does not use a dialog and does not directly use the computer facility of the present invention. Will be used to provide taste and preference APIs. For example, the preference and preference API allows the present invention to collect preference and preference information from users, provides user preferences and preference information to third parties from previously established preferences and preference profiles, It will allow parties to provide preferences and preference information for users based on the user's most recent actions. In embodiments, user preferences and preference profiles may be generated and updated from user interactions, responses, recommendations, reviews, etc., but the user need not interact with the facility of the present invention. In an embodiment, the system is known by using natural language processing, such as inferring preference profiles from likes and dislikes ratings or by analyzing how users have tagged user profiles. May learn user preference profiles. In this case, a similar user may have previously established preferences and preference profiles, and may have preferences and preference profiles that will match the user. In addition, these similar users may have reviews related to their profile. The system will match users similar to the user and provide reviews related to the user. For example, the user has an existing preference and preference profile using the system, either by interacting directly with a computer facility or via at least one third party API, and several products, services You might want to know what other similar users are thinking about, people, events, etc. The system may search for similar user preferences and preference profiles for subjects that the user is interested in. In this way, the system can provide such users with reviews from similar users, and will help them decide what they want to do based on their preferences and preferences. For example, a user may go to a merchandise website that uses the preferences and preferences API of the present invention and be interested in reviewing digital cameras. The third party will find similar users, search for digital cameras by those similar users, and provide reviews to the current user. In an embodiment, the review may be in a third party facility, another third party facility, a facility of the present invention, and the like. In embodiments, showing similar user reviews allows users to access related reviews more efficiently, and API third-party users are more targeted to their users. Could provide relevant support.

  In an embodiment, the present invention provides that a third party provides a review from a similar user to the user where the user is determined to be similar without the same user participating in the dialog via the present invention. It will provide the taste and taste API that you will use to do. For example, similar users may be identified as similar based on social networks, friends, family, work, etc. For example, the user is related to the first user via the social network, and from this relationship, “similar” may be determined depending on age, interest, and the like. Similar users will provide reviews of products and activities. This review will be provided to the user as relevant via similarity to other users. In an embodiment, similar users will be determined through similar recommendations to other topics, such as in combination with social relationships.

  In an embodiment, the present invention provides a computer program product recorded on a computer readable medium that, when executed on one or more computers, uses a computer facility to help a user find similar user reviews. Would provide. This provides (1) a user preference learning API to a third party website to determine user preferences when applied to a third party market, executed as an extension of a computer facility; ) Collect preferences of multiple users including users, (3) Store user preferences in preferences and preferences database, including multiple preferences and preference profiles, (4) Via third party preferences and preference learning API Receive a request from a user for a review of a topic from a user with similar preferences and preferences, and (5) match at least one of the user's preferences and preferences and preferences of other users in the preference database (6) Relevant requests for topic reviews from other matched users And (7) providing a review to the user. In an embodiment, determining the preference will be done using natural language processing. Reviews may be in computer facilities, third party facilities, etc. The preference learning API will enable at least one collection of cost information, product information, personal information, topic information, and the like. Reviews will be provided by similar users who do not have preferences and preference profiles where the user is determined to be similar from social relationships such as social networks.

  In an embodiment, the present invention will provide a preference and preference API that a third party would use to recommend goods, services, etc. For example, a user may come to a third party website looking for product recommendations, and the third party may use a preference and preference API to better understand what the user typically likes. Would use this preference to recommend products. For example, a user may come to an audio store website looking for recommendations about an audio system that plays music from an iPhone. The third party will use the user's existing preferences and preferences via the API. In this example, user preferences and preferences will indicate that they are college students and often enjoy social life. From this information, third parties will make recommendations, such as recommendations for portable, small, powerful, etc. audio systems. Alternatively, third parties may use preferences and preferences APIs to determine their preferences and preferences when querying users for targeted third-party content, general users, etc. . The third party will use this new preference and preference information either alone or in combination with previous preferences and preference profiles according to the present invention to make recommendations. The preferences and preferences established by third parties can also be used, combined with new preferences and preference profiles generated by other third party APIs, or directly by the facilities of the present invention. Will be stored in the facility. In embodiments, the ability to use preferences and preference APIs may be used to improve recommendations for goods, services, etc. made through third party sites.

  In an embodiment, the present invention will provide a preference and preference API that a third party would use to recommend products, services, etc. to the user based on similar user activity. For example, two users have previously established preferences and preference profiles of the present invention, one of the users has selected a product, service, etc., and a third party has It will make recommendations to other users based on their similarity, such as being determined. For example, two users will have similarities determined from their preferences and preferences by age, location, political opinion, social activity, and so on. The first user will select a product such as a car. If a second similar user must show interest in a car, such as by search, ad selection, explicit question to the present invention, etc., the present invention is potentially better because of their similarity As such, it would provide a second user with a car option. In an embodiment, this would be done without a dialog provided to one or both users.

  In an embodiment, the present invention will provide a computer program product recorded on a computer readable medium that, when executed on one or more computers, uses a computer facility to help users find recommendations. . This provides (1) a user preference learning API to third party websites to determine user preferences when applied to third party products and services, executed as an extension of computer facilities; (2) Receive third party information related to third party products and services, (3) collect user preferences, store information obtained from user interactions on the Internet as user preference profiles, (4 A) receiving a query from a user at a third-party website associated with at least one of the third-party goods and services; and (5) from the computing facility to the user based on the user's query and confirmed preferences. , At least one of goods and services Performing the step of providing a recommendation. In an embodiment, determining the preference will be done using natural language processing. The collection may be from at least one of third party websites on the Internet, multiple third party websites on the Internet, recommendations, purchases, and search result selections made by the user. The computing facility will be a machine learning facility. Third party information includes product information from the product manufacturer, product information from the website owner, price information from other websites, inventory information from other websites, price information from shop owners, inventory information from shop owners , Reviews, comments, ratings, etc. The API will enable collection of cost information, product information, personal information, topic information, and so forth. The preference will be derived from the behavior of the second similar user, and the similarity will be determined from the preferences and preference profiles for the user and the second similar user. A second similar user action would be at least one selection of goods and services.

  In an embodiment, the present invention will provide a preference and preference API that a third party social network site will use to show users similar to them on the social network. These similar people will be shown as a photo or as a list by region, age, gender, etc. For example, a user may request to come to a social network to meet and connect with people like them. Social network sites will use the preferences and preferences API to present users with dialogs that determine preferences and preferences for general or social situations, social networking, activities, music, personalities, and so on. Alternatively, the user already has a preference and preference profile determined by using the facilities of the present invention directly, or via other third party APIs, via social network sites, etc. There will be. Social network, other people on the social network, such as via information obtained about other people, such as obtained via social networks, via previously determined preferences and preference profiles of other people Will use this information to match the user. For example, user preferences and preferences would indicate that they are young and enjoy going to NYC clubs. Social networking sites will be able to match users with similar people on social networks by list or by photo, category, city area, etc. In an embodiment, a preference and preference API used with social networks could allow users who want to find other similar people with social experience to experience enhanced matching.

  In an embodiment, the present invention is recorded on a computer readable medium that, when run on one or more computers, uses a computer facility to help users find other similar users on social networking sites. Would provide a computer program product. This includes (1) receiving an initial request from a user via a third party social networking site API to find other similar users on the social network, and (2) via a social networking site API. The user's preference is confirmed, (3) the user's preference is matched with the other user on the social network who has the same preference, and (4) the user includes other users having the same user's preference. The step of presenting the matching result is performed. In an embodiment, confirming the preference will be done using natural language processing. Matching results will be presented to the user as a list of similar users. The matching results will be presented to the user as a similar user profile. The matching result will be presented to the user as a link to a similar user in the social network. The computing facility will be a machine learning facility.

  In an embodiment, the present invention provides a preference and preference API that a third party search facility would use to rank search results based on which results the same user selected the most. right. For example, the search facility will give the user a machine that improves the relationship in how search results are listed by preference and preference profile provided by the preference and preference API of the present invention. A preference and preference profile database, etc., will be aggregated and maintained for the search facility to rank user search results from previously selected results by other similar users. For example, a user may have a preference and preference profile that indicates a retired man who likes yachts and is somewhat adventurous. When users search for Caribbean vacation destinations, search facilities with these preferences and preference attributes are listed first, such as island yacht rental packages, island hikes, and strange places. Will rank. In an embodiment, the preference given to the search facility and the user of the preference API will improve the relevance of the ranked search results to the user.

  In an embodiment, the present invention provides a computer program product recorded on a computer readable medium that, when executed on one or more computers, uses a computer facility to help a user rank search results. right. This includes (1) receiving a search request from a user via a third-party search facility, (2) checking the user's preferences, and the confirmed preferences from the user are the user's preferences and preference profiles. Generated and stored in a preference and preference storage facility, including other user preferences and preference profiles, the profile also includes a history of search results selected by other users in previous searches, (3) Match users with other users who have preferences, (4) determine a set of search results for a user's search request, and (5) selected by other users with similar preferences and preference profiles Match the search result history with the set of search results and (6) selected by other users with similar preferences and preference profiles Follow the etching result, performing the steps of presenting to the user a ranked search results. In an embodiment, confirming the preference will be done using natural language processing. The computing facility will be a machine learning facility. The search facility will be a search engine.

  Referring to FIG. 34, the present invention is designed to infer unknown user preferences and preferences by finding routes via social interactive constructs based on the Internet to people with known preferences and preferences. Would use a social graph. Thus, the present invention provides a method for obtaining user data that the system did not know before. In an embodiment, the present invention, when executed on one or more computers, helps to determine unknown preferences using an internet-based social interactive graphic display on a computer facility 3402. A computer program product recorded on the medium will be provided. This is (1) a plurality of known users 3404, confirming the preferences of a plurality of users that are part of a social interactive construct based on the Internet, and (2) for a plurality of known users 3408, Determining an internet-based social interactive graphic display 3412; (3) based on the correlation between the unknown user and the plurality of known users in the graphic display 3410; A step of inferring an unknown user's preferences in the social interactive graphic display 3412 is performed. In an embodiment, the Internet-based social interactive graphic display may be a social network, a social graph, a social diagram, or the like. The unknown user will be three, five times, etc. away from the closest known user in a social interactive graphic display based on the Internet. The inferred preference of the unknown user will make the unknown user a new known user, and the new known user will be used to infer the second unknown user's preference. Preferences include personal information, topic information, etc. related to user interaction, and the interaction will be via a social interactive graphic display based on the Internet. The interaction will be via an API provided to a third party website. Inference will be provided in conjunction with other known users associated with the user in the social interactive construct, based on the user's Internet. Confirmation of preference will depend on the use of natural language processing. The computing facility will be a machine learning facility. Inferred preferences will be used to share reviews with unknown users to target ads to unknown users. Inferred preferences may be used to recommend products, services, etc. to unknown users. Inferred preferences will be used to help rank search results for unknown users. A known user in the vicinity of an unknown user will have a greater weight in the inference algorithm. Inferred preferences are refined by information from other sources, including third-party sources, recommendations made by multiple known users, search queries by multiple known users, multiple queries Search results selected by one of the known users, and personal preferences determined through web interaction by at least one of the plurality of known users. Other sources will include third party preference learning APIs.

  In an embodiment, the present invention provides a social network graph to infer unknown user preferences and preferences, or vice versa, by finding routes through social networks to people of known preferences. , Diagram, graphic display etc. will be used. A social diagram is generally a mapping between multiple users and how they relate. By using social diagrams, the preferences and preferences of known and unknown users will be determined from the interrelationships in the diagram. For example, a user with a known preference and preference profile will be directly related to multiple other users as displayed in the social diagram. In the first approximation, it will be assumed that these other users are similar to the user and have similar preferences and preferences. These other users will be provided with sophisticated services, taking advantage of knowing user preferences and preferences as described herein. For example, if there is a user with a known preference and preference profile indicating that he is a rock climber, the user in the first link of the user's social diagram is also assumed to be a rock climber. Let's go. In practice, this proves to be a too general process. However, the process by which the user is related to other rock climbers is good and the system will look for social diagrams to find other known users who enjoy rock climbing. In this example, it is found that there is another known user who is also 3 rock apart, who also enjoys rock climbing, and this user will be found to be in a cluster connected to the first user. From this, it would be a good assumption that this cluster is a group of rock climbers, and all rock climbers share a similar set of preferences and preferences. In an embodiment, preferences and preferences are inferred from associations in the social network diagram and will benefit from the invention described herein.

  In an embodiment, the present invention provides a computer readable medium that, when executed on one or more computers, helps determine the preferences and preferences of unknown users using a social network graphic display on a computer facility. Will provide a recorded computer program product. This is (1) confirms the user's preferences, being a known user, (2) determines the social network graphic display of the known user, and (3) other in the social network graphic display of the known user Determine the presence of a known user and (4) exist in the social network graphic display of the known user based on the correlation between the unknown user, the known user and other known users in the network graphic display The step of inferring the preference of an unknown user is performed. In embodiments, confirming the preference will be through the use of natural language processing. The social network graphic display may be a social graph, a social diagram, etc. The computing facility will be a machine learning facility.

  In an embodiment, the present invention will combine user preferences and preferences as determined by two or more third party APIs to improve recommendations provided by two or more third party APIs. For example, there are preferences and preference profiles created by one or more third-party APIs, and by combining these different preference and preference profiles according to the present invention, a combined preference and preference profile will be generated. Let's go. Furthermore, as additional preferences and preference profiles are generated by third party APIs, they will be used to continuously update the user's combined preferences and preference profiles. Third parties will use combined preferences and preference profiles to improve recommendations. This is especially true when different third parties focus on profiling preferences and preferences for different areas such as products, personal relationships, services, celebrities, and so on. It will be appreciated that combining several more specific profiles into a combined profile provides a richer taste and preference profile and can be generated from any one of the more specific profiles. In addition, users have recently interacted by changing preferences and preferences over time and combining more recent user profile interactions on one third-party API. Not, but will benefit other third parties who want to keep their users' profiles up to date.

  Referring to FIG. 35, in an embodiment, the invention is recorded on a computer readable medium that, when executed on one or more computers, provides improved preference and preference profiles through the use of a computer facility 3502. Would provide a computer program product. This is because (1) the first preference and preference profile of the user 3512 is generated by the interaction between the first third party website 3514 and the user via the first preference learning third party API 3504, and (2 ) Collect information on further user interaction with the second third party website 3514 via the second third party API 3508, and (3) preferences and preferences to improve preferences and preferences profile 3510 The step of combining the profile and the information of further user interaction is performed. In an embodiment, generating the first preference and preference profile would be by user preference confirmation through the use of natural language processing. The computing facility will be a machine learning facility. The API will allow collection of at least one of cost information, product information, personal information, and topic information.

  Referring to FIG. 36, in an embodiment, the graph construct 3620 will be developed and utilized by a web-based advice facility 3602. For example, it helps to provide recommendations to the user 3608 by dialog with the user 3608 via the Internet 3604 and provides recommendations to the user when the graph construct enhances the process leading to the recommendation. To do this, using a minimal dialog with the user and using a graph construct eliminates the need to continue the dialog with the user to make recommendations to the user, etc. In an embodiment, the graph construct will be constructed with information from a third party site 3610. In an embodiment, there may be multiple types of nodes 3612 in the graph, such as people, entities, tags, etc. For example, people may be users of websites, applications, mobile devices, shoppers, anonymous web browsers represented by unique cookie IDs, and the like. Entities may be things that people like, dislike, buy, search, study, etc. Tags may be short textual descriptions of entities, people, etc. In an embodiment, the nodes of the graph may be connected by multiple types of edges 3614, such as preference data, tag data, and the like. For example, an entity may be connected to a people node by a preference edge that represents the degree to which a person likes or dislikes the entity. Tags will be connected to people and entity nodes depending on how these people and entities are tagged with those tags.

  In embodiments, the graph data is given explicitly by a user (eg, user “A” says “B” likes), or by visiting a publicly available website, or This may be due to third-party sources. When data is received, the system will attempt to “alias” data that already exists in the system. For example, if the data says to the system that user “A” prefers restaurant “B”, the system matches restaurant name “B” by matching names, addresses, phone numbers, other information, etc. Try to identify what you already know about. This allows the system to aggregate data from all multiple sources for the same entity representing restaurant “B”, such as training data. The system will alias the user. For example, user jsmith99 may be the same user as john_smith on two different websites. The system will use similarities such as username, email address, photo, full name, geographic location, etc. to associate users and identify systems across different websites.

  In an embodiment, the nodes of the graph will have a “preference profile”, such as having a numerical value. One would be expected to like or dislike an entity or tag based on their preference profile. Similarly, two people would be expected to be similar or not similar based on their preference profile. Entities will also be compared to see how similar they are to each other using these preference profiles. The graph of the system initially has preference profiles assigned to certain nodes and propagates those preference profiles to nodes that do not have preference profiles. This propagation would be a recursive process that “flows” a preference profile from a node that has a profile to a node that does not have a profile. Alternatively, the recursive process will update the profile of a node that already has a profile based on the profile of the adjacent node. New data will be merged into the graph by adding new nodes or edges and simply using the adjacent nodes to update the new or changed nodes. Alternatively, the system will run multiple restarts of updates for the entire graph.

  Many different types of data will be fitted so that they can be viewed as “likes” or “dislikes”. For example, viewing a web page can be graphed as a weakly coupled edge between the person viewing the web page and the entity representing the web page. A person who buys a book can be represented by an edge that makes a strong bond between the person who buys the book and the book itself. A person answering a question with three mutually exclusive answers can be represented by the person, the edge between the answered answers, and two negative edges to the two answers that the user did not answer.

  In an embodiment, the method and system will provide recommendations to the user based on the degree of recommendation, here called “degree of interest,” whether the recommendation is new or interesting. In an embodiment, the degree of interest may be a combination of a subject matter, topic, or product that is of interest and how new or revised the idea is. In an example, the user lives in the United States and is interested in making Italian food, and the system will therefore recommend cooking to the user. In this example, as a Parmesan cheese, it is recommended to try the taste of “new” cheese, because using Parmesan cheese is not new in the United States, but rather a lot of interest, Would have. Alternatively, pecorino cheese is an Italian cheese made from sheep's milk, which can be used in pasta dishes instead of parmesan cheese and is often preferred if a sharp taste is desired. Thus, the recommendation to replace pecorino instead of parmesan would be considered to have a high degree of interest, at least relative to that of parmesan. Interest levels may be determined relative to standard or typical, relative to past recommendations, relative to new factors, and so on. In this example, Pecorino's interest would be relatively high compared to Parmesan.

  In embodiments, interest is related to a known preference profile and is fresh to the user, which is not new (such as new all over the world), the user's past experience ( It may mean that the user's known history is referred to). Alternatively, “fresh” may be new facts or stories related to something of new interest that is not necessarily new to the world or users. For example, “Chipotle on 21st Street” is neither new nor interesting, but if you recommend that Oprah buys a burrito with a chipotre that buys a burrito, then it ’s interesting. It is. The system recommends this by requiring the user to write a reason for recommending it. The degree of interest will be determined in relation to the ratio of the user's item rating to the item stored by the user. For example, if there are facilities such as bookmarks, save later, add to wish list, etc. (eg star rating facility), between items with a high rate, which are “low saved” items as not of interest You can see that it is related. In this example, everyone knows about the item and can rate it, but it is not worth saving for later and may not be of interest. Alternatively, items that are highly rated and often stored by the user are considered interesting. This is because items are considered high (ie, highly rated) and are worth saving for further consideration. The degree of interest will be the acceleration of social activities. For example, there are many more foursquare. There will be a restaurant where you can get com check-ins. This is a sign that something interesting is happening there. In an embodiment, the advisory facility also looks at the release date of books, movies, albums, products, etc. that are considered to be new in nature, sees the date the first review was written on the web, releases the movie The earliest date that the item can be found on the Internet in order to determine what is new or new to the user and determine their interest, such as by viewing events such as concerts, author performances, etc. Will adopt. The advice facility asks the user to rate what the user already knows, and the system will assume that items that the user did not rate are new to the user. The advice facility will prompt the user to use the “Save” feature if the user does not know it yet but wants to check it.

  Thus, “interest”, as used herein, includes relevance in the traditional sense (more specifically, quantitatively assessed by relevance), and in particular, the user's profile (or , Preferences profile) and the relationship between new content. A wide variety of analytical, mathematical, rule-based and / or heuristic techniques are known for assessing relevance, any of which, as mentioned here, It may be useful to determine relevance (more generally, interest). However, the degree of interest further includes a dynamic relationship between the user and the content based on, for example, time, location, user history, and the like.

  For example, time may simply be important as a measure of freshness, such as whether current statistics are more interesting than older measurements of the same statistics. Conversely, at certain points in time, including historical times (as determined by metadata, content, chronology, etc.), or historical periods, old statistics, facts, opinions, etc. Showing interest will be more interesting. Newness, that is, how recent an item is, there are many different items of information about a topic, and some degree in the current popularity of or interest in a particular thing in various items This is especially important for increased interest. This type of popularity is based on activity on blogs, passive measurement of newly-incorporated web content, or other Internet-based measurements of user interest, such as hits, traffic, or web servers. It can be measured in a number of ways, including other activities, active measurements such as client behavior groups or individual monitoring. The time is also important for the item's interest in other contexts, such as when the time is explicit or implicit in the user's query, such as what to do this week, a movie tonight, etc.

  The location will also have a significant impact on interest. This includes simple, geographic neighborhoods using appropriate location-finding techniques, and other aspects of the user profile, such as interest in a particular location (eg cooking, art, entertainment) or current activity associated with the user. Will include aspects. However, inferences about the convenience of neighboring locations using various means of transportation available to the user (eg cars, public transport, etc.) and possible user budgets for immediate or long-term travel planning It will be understood that it will also include items related to the location. In addition, the location and the context of interest associated with it will depend on other dynamic location attributes, such as the location of friends on the social network, the proximity and distance to the geographical location of the friends.

  User history may also be used to parameterize interest. For example, if a new item corresponds to the user's query or matches the user profile well, but differs significantly from the previous content acquired by the user, the difference is that the calculated relevance is The item will be more quantitatively interesting even if it is less than the relationship of other outcomes. Thus, in one aspect, the degree of interest is simultaneously dependent on similarity (or relevance, etc.) and dissimilarity, or a measure of the feature that makes an item dissimilar to previous content in the user history. right. Alternatively, items of information may not rank well based on a general measure of relevance that is modest, based on other aspects of the user's current context. Therefore, the degree of interest is explicitly measured based on dissimilarity of information in the user's history or on an objective basis that employs relevance based on the user's context, Newness for the user will provide the user with a relevant measure based on an appropriate similarity or matching metric that is further emphasized. In one aspect, interest is objectively measured as relevance based on the user profile, along with the user's history and dissimilarity to one or more aspects of the user's current context, such as time and place. right. By being measured in this way, many objectively relevant items will not be of particular interest to the user, and somewhat related items will be of great interest.

  Referring to FIG. 37, recommendations may be provided via the recommendation facility 3704 as part of a web-based advice facility 3702. In an embodiment, the recommendation facility will utilize interest filtering 3708 in the process of generating recommendations for user 3722. Recommended sources of information may include friends 3718, similar users 3714, influential people 3720, information source website 3712, and the like. The recommendation may be provided to the user via the Internet 3710, via a telephone facility 3724 (eg, a cellular network), and so on.

  In an embodiment, the interest level determination may include other individuals' “social activities” (eg, friends, celebrities, authorities), products, places, and “born” dates of events (eg, restaurant opening dates, movies) , New product release date) and so on. Other personal social activities include friends with high topic ratings, high topic ratings, similar favorite friends, influential non-friends with high topic ratings, as well as high ratings for this topic Individuals who are influential to the user, such as non-friends of special taste, may be involved. These other individuals are friends, people who don't necessarily know, but people who show similar preferences in this topic, people who don't necessarily have the same preferences as users, but are well-known, wealthy, well-known people Would fall into different categories, such as critics of this topic (eg movie critics). The system will also be aware of which user's friends have similar preferences in this particular topic. For example, if Ted has a restaurant preference similar to the user, but Alice is not, the restaurant is of interest to the user if Ted likes it, If you like, you will not necessarily be interested. Other reasons why something might be of interest to the user are when a friend later saves a recommendation, or when a friend is actively commenting or discussing a recommendation.

  In an embodiment, the recommendation is based on the current geographical location with respect to the search topic, periodically, as a data feed, push service, via email, social network, third party site, if the user requests. , Will be supplied to users in connection with home computers, mobile computers, mobile communication facilities (eg, mobile phones, smartphones, PDAs), etc. For example, a user may select a mobile phone based on the current geographical location, such as a product recommendation for a store in the area (eg, a sale product, a new product, a product that is difficult to obtain), a place to see, a restaurant to try, etc. An interest recommendation is provided, and the recommendation will be based on the interest. In this way, users not only receive recommendations, but rather “find” new ideas, places, products, etc. Will receive. And if discovery is related to a particular genre, it could be “local discovery”, “restaurant discovery”, “technology discovery”, “dishes discovery”, etc.

  In an embodiment, the present invention provides a user with “local discovery”, which “types” something in the search box or “brings” the search results. Rather than relying on it, it provides users with something new and interesting. This is more limited for portable devices that are more difficult to type when device input is limited (eg, small keyboard, small display, user moving (walking, driving), etc.) Will be useful. The description of local discovery here is mainly about application to mobile devices, but those skilled in the art can implement it on any computing facility such as laptops, desktops, navigation devices, etc. Will be understood. Local discovery facilities are obtained via a web interface, via a “new” email for the user, via Twitter, via a block platform (eg, Wordpress, Thumbr, etc.), etc. You can also. In addition, local discovery content will be provided to the user upon request, forwarded to the user (eg, email), pushed to the user for new interest (eg, weekly), and so on.

  In an embodiment, the application of local discovery of a mobile device is an authoritative, rated in the application in the past, rated by their friends, rated by people with similar preferences to the user, Shows users nearby places, such as a source rated by a source, a celebrity rated, etc., and a "rating" is an inference that is inferred from human behavior (eg, online or offline) Let's go. In embodiments, the locations shown may be restaurants, bars, boutiques, hotels, etc. There may also be navigation elements that allow users to filter the list down to a specific location, such as a nearby “Italian restaurant” or a specific location recommended by others.

  Instead of indicating the location, local discovery will also indicate to the user what to buy, what event to go to, what to see (if you want to kill time), and so on. For example, local discovery may provide a list of recommended books that a user may want to read and optionally indicate where to buy it nearby. Local discoveries include what the user's friends have recently liked, what people with similar book preferences have recently liked, what popular / great critics have recently liked, or They may simply select items and provide recommendations based on machine learning, such as what is popular overall or what is popular near the user. Similarly, this also applies to other types of goods, events to go, etc.

  In embodiments, local discovery will find similar favorite people in each area (restaurants, books, etc.) and let the user follow what they have rated. For mobile applications, local discovery will use the user's location to filter and narrow the list of nearby, similar people like the user. Local discovery determines, via machine learning, whether other people have similar preferences to the user, in which case they ask both to rate various places and things, You might ask them to answer questions to measure similarity. Local discoveries, for example, by advising facilities, prefer to be vague together, dislike popular ones, show reviews written by others, and other people's characteristics (demographics, places, etc.) ), Or how many other people are following other users, and so on, will try to see the similarity between the user and others.

  In an embodiment, rather than generating a list of recommended places, things, and events for a region, local discovery also provides a “discovery” of what is interesting to the user, such as having a high interest rating Will generate. This would mean that every time a user searches for a place close to the office, he sees several different results every day, rather than seeing the same 10 restaurant recommendations. Ideally, these new restaurants will be displayed based on similar preferences, user friends or people who prefer new places that are nearby, and these will also be staff members who implement local discovery at all times It may also be partly an editing process that discovers new places and sends them to the user's device.

  In an embodiment, the user will also save to a “wish list”, “to do list”, etc. for later use. This saved list of products, places, events, etc. also about these products, places, events, deals, availability, new reviews, etc. so as to draw attention only to the user's interest Would be used to draw attention to the user. For example, if a user's phone vibrates to tell you that you can buy some shoes in a store that is about to pass, if you didn't have previously shown in the Wish List that you want those shoes, It may be frustrating.

  In embodiments, the user interface for local discovery applications is a map, text list, “cover flow” (such as an Apple computer cover flow implementation) such as a flip-to-turn interface, etc. Let's go. The interface also provides a reminder to the user when a friend or similar favorite likes something nearby or likes something that the user has stored in the wish list. Would do.

  In an embodiment, the local discovery application is related to the weather, and the application will determine the recommendation in part based on the actual or expected weather in the area where the recommendation is desired. For example, if the weather is warm and it is not raining, the application recommends a very good place to sit outdoors, if the weather is good, recommend something for enjoying the outdoors, and if the weather is bad Would be weighted by a recommendation to go to the museum.

  In an embodiment, the local discovery application may relate to the time of day, such as taking into account the time of activity, the distance to the location when making recommendations. For example, if you now want a place to eat, the system will not recommend places that are closed or likely to close. Similarly, the system will not proceed to do things that have already been sold out, will not arrive in time given the estimated travel time, and so on.

  In an embodiment, the local discovery application may be related to preferences, such as using the person's preferences when recommending, rather than simply showing what is popular among nearby users. In embodiments, preferences may be inferred from what the user likes, people who follow on social networks, etc., as described herein.

  In an embodiment, the local discovery application uses the location of the person when recommending or uses their location when providing the user with a list of topics that can be recommended. It will be related. For example, if there is no hair salon nearby, the system will not provide the user with the option to obtain a hair salon recommendation. Similarly, if there are no Mexican restaurants or video game stores, the system will not offer a recommendation choice for these topics. If the user's location is in the store, the system will not provide recommendations on topics that the store is not keeping in stock or selling.

  In an embodiment, the local discovery application is related to social activities, such as showing recommendations that have received a certain area of activity from friends, authoritatives respecting the topic, people like the user, etc. Will do.

  In an embodiment, the local discovery application, as described herein with respect to interest, is new, newly popular, received social activity, is not new, It may be related to a recommendation that interests you, such as there is a sign that is not so popular that you are not likely to know it.

  In an embodiment, a local discovery application may be a recommendation that a person is interested in using a mobile application, what they see in a store, what they see while taking a walk, and so on. , Would be related to items stored by the user on the mobile device. For example, when you find a book you like at a bookstore, you might scan the barcode and save it for later. When you see a restaurant you like, you'll take a photo, capture the location with GPS, and save it for later. Stored content will be the source of recommendations for use in later mobile applications, and content will be sent to users via other channels, such as weekly email reminders, via web applications, etc. . Save will also be utilized by the system as a social indicator, such as showing other users that something is interesting content.

  In an embodiment, the local discovery application may relate to what items the user has stored on the Internet. This is similar to saving to a mobile device, where the user will save the content found on the web for later use in a mobile application. For example, if a user sees a book review and saves it for later, it will be reminded about it later when looking for a book recommendation using a mobile device. Similarly, restaurants and things to do will be saved and later reminded about it on mobile devices.

  In an embodiment, the local discovery application displays a topic that is to provide recommendations, such as displaying a grid of photos when the mobile application first begins to represent a topic from which recommendations can be obtained. It will be related. The choices are based on the user's location, application usage history, etc., and the application will predict what the user is interested in. For example, when an application is started based on the app's belief that a restaurant close to the user is recommended and the user is interested in the iPhone app and video game, among other things, the restaurant, iPhone app, video game There may be a photo. Actual photos will be picked up based on the user's knowledge of the application. For example, a restaurant topic photo may be a photo of a nearby restaurant that the application predicts that the user will prefer. The size and arrangement of the photos shows how interesting the user is in that topic, as shown in Figure 38, so that the topic most commonly used by the user is at the top of the list and is the largest photo. Or will be based on what the application predicts. The user would select one of the recommendations and be linked to more detailed information related to the recommendation, as shown in FIG.

  In an embodiment, the recommendations are as in an irregular grid where the system arranges from left to right and top to bottom based on how interested the user is in each recommendation and topic. Will display the image in an array as a function of the user's interests. For example, as shown in FIG. 40, the system predicts that the user is more interested in the restaurant than the coffee shop, and the restaurant tile is listed before the coffee shop tile. Also, the content of each tile is a recommendation on that topic that the user will like. Thus, restaurant tiles indicate restaurants that a nearby user would like, movie tiles display movies that would like music, and so on.

  In an embodiment, the local discovery application is concerned with determining the radius from which the recommendation is obtained such that the application must pick up the radius of how far the recommendation is from the user's location. right. The radius will be selected by the user, such as based on the population density of the area around the user. For example, in NYC, the radius may be 0.025 miles, whereas in a rural SC, the radius may be 60 miles.

  In an embodiment, when a local discovery application displays a recommendation to the application, which people should see the reviews from it, as it would be accompanied by reviews, ratings, or other recommendations from people It will be related to the decision. As shown in FIG. 41, the application selects which people to show based on whether it has the same preference as the person using the application, whether it is a friend of the user, an authoritative critic, etc. right.

  Referring to FIG. 42, the method and system will provide recommendations to the user through a computer-based advice facility 4202. Then, the topic information is collected, and the collected topic information includes the aspect related to the degree of interest of the topic and the aspect of interest, and the collected topic based on the interest aspect 4202 Information is filtered, and a computer-based advice facility 4208 is used to determine an interest rating from the collected topic information, and based on the interest rating 4210, recommendations related to the topic information are made. Will provide to users. In an embodiment, the aspect of interest could be derived at least in part from other individual social activities that show recommendations for the topic. Other individuals may be friends, celebrities, authoritative people, etc. Other individuals may have similar preferences as the user, generally or for a particular category or type of interest. A social activity would be to save recommendations. Social activity will be at least one of commenting and actively discussing recommendations. Social activities will be collected from other individual activities against sources on the Internet, such as social networking activities. The interest aspect of the topic information will also be determined based on an indication that the topic information has some aspect of freshness or novelty. The novelty will be an indication that the topic information is information on a new topic in a given period of time. The novelty will be a new popular sign of topic information. The new popular status will be determined by the level of activity on the web. Newness is common, such as the emergence of a new popular topic within a social network, or the old topic is exposed to the user for the first time, even if it is not the whole social network, and is new to the user , Will be specific to the user. The interest aspect of the topic's information is based at least in part on at least one of reviews, recommendations, blog entries, tweets, authoritative sources, news sources, electronic documents, purchasing, browsing, viewing time, etc. There will be. Without limitation, the aspect of interest will be based on time data. The time data will be the release date of the movie, product, etc. The time data may be the opening of an event such as opening a restaurant or opening a cultural event. An aspect of interest may be frequency data, such as how often topic information is referenced in online sources. The aspect of interest may be related to the interaction between the computer device and the user. User interactions will be interpreted by the machine learning facility as user behavior indicating the user's preference level of topical information. The user interaction will be the selection of a web link. User interaction may be at least one of tapping, touching, or clicking on the screen of the computing device. Computer-based advice facilities will include machine learning facilities. A computer-based advice facility will include a recommendation facility. Filtering will include collaborative filtering. The recommendation will be sent to the user's mobile communication facility to provide a recommendation in the user's current geographic region. These would be the graphic user interface of the mobile communication facility that provides the user with the ability to refine the recommendations provided to the user. The recommendation will indicate to the user at least one of the items to buy, the events to go, what to see, etc. The recommendation will be related to the local store. The recommendation would be related to local restaurants. The recommendation will be related to the local bar. The recommendation will be related to entertainment. The recommendations will be further filtered for the user based on their local area specific interest. The provision of recommendations will be sent to the user for the current geographic region in which the user is present. Recommendations will only be sent to users who meet the threshold of confidence and the system how much the user will like recommendations. The confidence threshold may be related to the rating of interest. The confidence threshold will be determined by a machine learning facility based on the user's past behavior associated with recommendations previously provided by the system. The user will be able to save the recommendation in the mobile communication facility storage. The user would be able to save the recommendation to a computer based advice facility storage.

  Referring to FIG. 43, the method and system uses a computer-based advice facility 4302 to provide a user with a geographically localized recommendation and recommend a recommendation determined to have an aspect of interest. Would collect from internet sources. The recommendation will be further determined based on the geographic location aspect 4304. In addition, the option compares the collected recommendations with the derived user preferences and the user's current geographic location 4308 to determine at least one recommendation to the user based on the processing for the comparison 4310, and Distributing at least one recommendation to the mobile communication device, wherein the user at least partially sees and saves the recommendation, such as through an application resident in a computer-based advice facility 4312, One of the sharing would be possible.

  In an embodiment, the computer-based advice facility would be a mobile communication device. The mobile communication device will be a smartphone. Viewing will include providing source information from Internet sources. Source information may include original recommendations, ratings, images related to Internet sources, and so on. The image may be a picture of the individual who provided the recommendation, or other indicia such as an icon representing such an individual. The source information will include a visual indicator of the degree of similarity of the recommendation of at least one person who made the recommendation. The indicia may be qualitative (eg, “this individual has a very similar preference to you”) or indicate a measure of relative similarity (eg, “you are Sharing 10 interests of the category "), will be quantitative. The source information will include a visual indicator of the degree of preference similarity over the recommended individual. A visual indicator will show a sort of individuals based on the degree of preference similarity. Individuals will be listed in order of decreasing similarity.

  In an embodiment, the interest aspect is information on a new topic, as determined by the date the information appeared in a domain, such as an Internet domain, a collection of Internet news sources, a corporate network, a social network, or the entire Internet. there will be.

  Interest aspects can be traced back to the user's browser history, as reflected by tracking the user's activity on one or more devices, such as reviewed by the user, accessed by the user, etc. It would be information on topics that the advice facility considers new to the user compared to content.

  In an embodiment, the aspect of interest is to a new item that the user has shown interest in the past, as reflected in the user's activity, such as user feedback, access to the item, time spent reviewing, etc. Information on topics that have new aspects to existing topics, such as updates.

  The aspect of interest will be determined in relation to the user's rating of the item or the ratio to the item stored by the user.

  In an embodiment, the aspect of interest will be determined by the acceleration of social activity related to topical information.

  In an embodiment, the aspect of interest is similar to the user profile (e.g., matching a category of the user's interest that the user has shown interest in the past and relevance to other users with similar preferences). Or similar to the item, such as showing interest), but at the same time, the user's history is not similar (eg, determined to be new to this user in one of the above) It will be determined based on information having the above aspects. Thus, for example, a user who has shown interest in a particular celebrity in the past may be expected to be very interested in new breaking news items about that celebrity.

  In embodiments, the aspect of interest is also new to the user's current location and domain (including the entire Internet, but optionally based on what is new for the domain linked to the user's current location), etc. Based on information on new topics determined by one of the factors described in this disclosure, based on information on topics considered new to the user by an advice facility, determined in relation to the ratio of the user's rated items Based on information about topics that have new aspects of existing topics, determined by acceleration of social activities related to topic information, based on temporary factors such as based on items saved by the user Will be decided. The geographical aspect is the geographical location associated with the topic information, such as the location of the event, the location of the store, the location of the restaurant, the location of interest, the location of at least one product, etc. Let's go.

  In various embodiments, derived user preferences are based on ratings, which are inferred from the user, the user's friends, people like the user, authoritative sources, celebrities, and user behavior. Or machine learning about the user's online behavior. User behavior may be online behavior, including shopping behavior, browsing behavior, social networking behavior, location behavior, and so on. Recommendations may be items to buy, places to visit, events to participate, places to eat, and the like. The recommendation will be based on one of the rating and recommendation of at least one other user that is similar to the user's preference from the user's current geographic location. Similar preferences will be determined by machine learning via ratings from other users and / or other users' online behavior. Recommendations will be supplied as part of the provision of local discovery recommendations. A recommendation is a save that saves a recommendation in a list, which may be a wish list, a to-do list, an event list, a transaction list, and so on. The saved recommendation will be shown to the user via the local discovery application if the geographic location aspect of the saved recommendation matches the user's current location. A recommendation is sent to the user based on a recommendation from at least one other user, and the advice facility will determine the applicable radius around the user's current location in using the other user's recommendation. . The application is a local discovery application, which associates at least one of the new and stored recommendations for the weather, and the new and stored recommendations for the time of day. Associate at least one and associate at least one of the new recommendations with a stored recommendation about the user's social activities. The local discovery application will display an image based on how much the advice facility considers the user is interested in each recommendation and at least one of each topic. The displayed images are irregular, with the order from left to right, top to bottom, the advice facility considers how much the user is interested in each recommendation and at least one of each topic. Will appear in the grid.

  The methods and systems described herein may be deployed, in part or in whole, via computer software, program code, and / or machines that execute instructions to a processor. The present invention may be implemented as a method on a machine, a part or related system or apparatus of a machine, or a computer program product stored on a computer readable medium running on one or more machines. . The processor may be part of a server, client, network infrastructure, mobile computing platform, fixed computing platform, or other computing platform. A processor may be any kind of computing or processing unit capable of executing program instructions, code, binary instructions, and the like. A processor may be a signal processor, digital processor, embedded processor, microprocessor, or any coprocessor type (math coprocessor) that executes stored program code or program instructions directly or indirectly. Graphic coprocessor, communication coprocessor, etc.) or will include. In addition, the processor will allow execution of multiple programs, threads, and code. Threads will be executed concurrently to increase processor performance and to facilitate concurrent execution of applications. Depending on the implementation, the methods, program code, program instructions, etc. described herein may be implemented in one or more threads. The thread will spawn other threads that have given priority. The processor will execute these threads based on priority or any other order based on instructions provided by the program code. The processor may include memory as described herein or for storing other methods, code, instructions and programs. The processor will access the storage medium through an interface described herein, or other method, code, and instruction storage. A storage medium associated with a processor for storing methods, programs, code, program instructions, or other types of instructions executable by a computing or processing device is one or more CD-ROM, DVD, memory , Hard disk, flash drive, RAM, ROM, cache and the like.

  The processor will include one or more cores that enhance the speed and performance of the multiprocessor. In an embodiment, the process would be a dual core processor, a quad core processor, other chip level multiprocessors, etc. that combine two or more independent cores (called dies).

  The methods and systems described herein may, in part or in whole, execute servers, clients, firewalls, gateways, hubs, routers, or computer software on other computers and / or networking hardware. Will be deployed through the machine. The software program may be related to a server including a file server, a print server, a domain server, an Internet server, an intranet server, and other modified devices such as a second server, a host server, and a distributed server. A server may be connected to one or more memories, processors, computer readable media, storage media, ports (physical and virtual), communication devices, other servers, clients, machines, devices via wired or wireless media. It will include an interface that can be accessed. The other methods, programs, or code described herein will be executed by the server. Furthermore, other equipment required to perform the methods described in this application would be considered part of the infrastructure associated with the server.

  The server will provide an interface to other devices including, but not limited to, clients, other servers, printers, database servers, print servers, file servers, communication servers, distributed servers, and the like. Further, this coupling and / or connection will allow remote execution of programs across the network. Some or all of these devices will allow parallel processing of programs or methods in one or more locations without departing from the scope of the present invention. In addition, any device connected to the server via the interface will include at least one storage medium capable of storing methods, programs, and / or instructions. The central repository will provide program instructions to be executed on different devices. In this implementation, the remote repository will function as a storage medium for program code, instructions, and programs.

  The software program may be associated with clients including file clients, print clients, domain clients, Internet clients, intranet clients, other variants such as second clients, host clients, distributed clients. A client can communicate with one or more other clients, servers, machines, devices via memory, processor, computer-readable media, storage media, ports (physical and virtual), communication devices, wired or wireless media Will include accessible interfaces etc. The other methods, programs, or code described herein will be executed on the client. Furthermore, other equipment required to perform the methods described in this application would be considered part of the infrastructure associated with the client.

  A client will provide an interface to other devices, including but not limited to servers, other clients, printers, database servers, print servers, file servers, communication servers, distributed servers, and the like. Further, this coupling and / or connection will allow remote execution of programs across the network. Some or all of these devices will allow parallel processing of programs or methods in one or more locations without departing from the scope of the present invention. Further, any device connected to a client via an interface will include at least one storage medium capable of storing methods, programs, applications, code and / or instructions. The central repository will provide program instructions to be executed on different devices. In this implementation, the remote repository will function as a storage medium for program code, instructions, and programs.

  The methods and systems described herein may be deployed in part or in whole via a network infrastructure. Network infrastructure is known in the art, such as computing devices, servers, routers, hubs, firewalls, clients, personal computers, communications devices, routing devices, other active and passive devices, modules and components, etc. Will contain the element. Computing and / or non-computing devices associated with the network infrastructure will include storage media such as flash memory, buffers, stacks, RAM, ROM, etc., apart from other components. The processes, methods, program codes, and instructions described and other herein will be executed by one or more network infrastructure elements.

  The methods, program codes, and instructions described herein and others will be implemented in a cellular network having multiple cells. The cellular network may be a frequency division multiple access (FDMA) network or a code division multiple access (CDMA) network. A cellular network will include mobile devices, cell sites, base stations, repeaters, antennas, towers, and the like. The cellular network may be GSM, GPRS, 3G, EVDO, mesh, or other network type.

  The methods, program code, and instructions described herein and other will be implemented on or through the portable device. Portable devices will include navigation devices, mobile phones, mobile phones, mobile personal digital assistants, notebook PCs, palmtops, netbooks, pagers, e-book readers, music players, and the like. These devices, apart from other components, may include storage media such as flash memory, buffers, RAM, ROM, one or more computing devices. A computing device associated with a portable device will be capable of executing stored program code, methods, and instructions. Alternatively, the portable device may be configured to execute instructions in cooperation with other devices. The portable device will be configured to communicate with the base station that interfaces with the server and execute the program code. Mobile devices will communicate over peer-to-peer networks, mesh networks, and other communication networks. The program code may be stored in a storage medium associated with the server and executed by a computing device embedded in the server. A base station will include a computing device and a storage medium. The storage medium will store program code and instructions that are executed by a computing device associated with the base station.

  Computer software, program code, and / or instructions may be stored and / or accessed on a machine-readable medium. Machine-readable media are computer parts, devices, recording media that holds digital data for a given period of time, semiconductor storage known as random access memory (RAM), optical discs, hard disks, tapes, drums, cards , And other types of magnetic storage forms, processor registers, cache memory, volatile memory, non-volatile memory, optical storage such as CD, DVD, flash memory (eg USB stick or key), floppy Disk, magnetic tape, paper tape, punch card, stand-alone RAM disk, zip drive, removable mass storage, offline removable media, dynamic memory, static memory, read / write storage, mu Will include other computer memory such as Bullable Storage, Read Only, Random Access, Sequential Access, Location Addressable, File Addressable, Content Addressable, Network Attached Storage, Storage Area Network, Barcode, Magnetic Ink .

  The methods and systems described herein will convert physical and / or intangible items from one state to another. The methods and systems described herein will also convert data representing physical and / or intangible items from one state to another.

  Elements described and illustrated herein, including flowcharts and block diagrams across the drawings, indicate logical boundaries between the elements. However, in accordance with software or hardware engineering practices, the described elements and their functions are represented as a monolithic software structure via a computer-executable medium that enables a processor to execute stored program instructions. Implemented on a machine as a single software module or as an external routine, code, service, etc., or any combination thereof, all such implementations are within the scope of this disclosure. Examples of such machines include but are not limited to personal digital assistants, laptops, personal computers, mobile phones, other portable computing devices, medical devices, wired or wireless communication devices, transducers, chips Computer, satellite, tablet PC, e-book, gadget, electronic device, artificial intelligence device, computing device, networking device, server, router, etc. Further, the elements shown in the flowcharts and block diagrams, or any other logical component, may be implemented on a machine capable of executing program instructions. Thus, the preceding figures and descriptions describe the functional aspects of the disclosed system, but if explicitly stated or not obvious from the context, the software for implementing these functional aspects A specific configuration should not be derived. Similarly, it will be understood that the various steps identified and described above will be altered and the order of the steps will be adapted to the particular application of the disclosed technology. All such variations and modifications are intended to be within the scope of this disclosure. Accordingly, a description and / or description of the order of the various steps is required in a particular application, or is explicitly stated or, if not apparent from the context, a particular order of execution of those steps. It should not be understood to require.

  The methods and / or processes and steps described above may be implemented by hardware, software, or any combination of hardware and software suitable for a particular application. The hardware may include a general purpose computer and / or a dedicated computing device, or a special computing device, or a particular aspect or part of a particular computing device. Processing may be implemented by one or more microprocessors, microcontrollers, embedded microcontrollers, programmable digital signal processors, or other programmable devices, along with internal and / or external memory. Processing can also or may not be for special integrated circuits for applications, programmable gate arrays, programmable array logic, or any other device or device configured to process electronic signals. Will be embedded in the combination. It will be further understood that one or more processes may be implemented as computer-executable code that can be executed on a machine-readable medium.

  Computer-executable code is stored for execution by the apparatus and processors, heterogeneous combinations of processor architectures, or different hardware and software combinations, any other machine capable of executing program instructions, Compiled and interpreted structured programming languages such as C, object oriented programming languages such as C ++, or any other high or low level programming language (such as assembly language, hardware description language, database programming language, etc.) Will be generated using technology).

  Thus, in one aspect, each of the methods described herein and combinations thereof will be embedded in computer-executable code that performs those steps when executed on one or more computing devices. In other aspects, the method is embedded in the system performing the steps and distributed among the devices in various ways, or all functions are integrated into a dedicated, stand-alone device or other hardware. It will be. In other aspects, the means for performing the steps associated with the processes described above may include any of the hardware and / or software described above. All such permutations and combinations are intended to be within the scope of this disclosure.

  Although the present invention has been disclosed in the context of the preferred embodiment shown and described in detail, various modifications and improvements will become readily apparent to those skilled in the art. Accordingly, the scope of the invention is not limited to the examples described above, but should be understood in the broadest sense permitted by law.

  All documents referred to here are merged here as references.

Claims (52)

A method of providing geographically localized recommendations to a user through a computer-based advice facility, comprising:
Collect recommendations from internet sources that are determined to have an interest aspect and a geographic location aspect;
Comparing the collected recommendations with the derived user preferences and the user's current geographic location;
Determining at least one recommendation to the user based on the comparison process;
Distribute at least one recommendation to the user's mobile communications device;
The method wherein the user is allowed at least one to view, store, and share the recommendation by an application that is at least partially resident in the computer-based advice facility.   The method of claim 1, wherein the computer-based advice facility is a mobile communication device.   The method of claim 2, wherein the mobile communication device is a smartphone.   The method of claim 1, wherein the viewing includes providing source information from the Internet source.   The method of claim 4, wherein the source information includes an original recommendation.   The method of claim 4, wherein the source information includes a rating.   5. The method of claim 4, wherein the source information includes images associated with the Internet information source.   8. The method of claim 7, wherein the image is a picture of an individual who provided the recommendation.   5. The method of claim 4, wherein the source information includes a visual indicator of a degree of preference similarity of at least one person who made the recommendation.   5. The method of claim 4, wherein the source information includes a visual indicator of the degree of preference similarity over the person making the recommendation.   The method of claim 10, wherein the visual indicator indicates sorting individuals based on a degree of preference similarity.   The method of claim 11, wherein individuals are listed in an order of decreasing similarity.   2. The method of claim 1, wherein the degree of interest is information on a new topic that is determined by a date of appearance on the Internet.   The method of claim 1, wherein the interest aspect is information on a topic considered new to the user by the advice facility.   The method according to claim 1, wherein the aspect of interest is information on a topic having a new aspect with respect to an existing topic.   The method of claim 1, wherein the interest aspect is determined in relation to a ratio of an item rated by the user to an item stored by the user.   The method of claim 1, wherein the interest aspect is determined by acceleration of social activity associated with the topic information.   The method of claim 1, wherein the aspect of interest is information that resembles a user's profile and is determined not to resemble a user's history.   The method of claim 18, wherein the degree of interest aspect is further determined to be at least one of a user's current location and a temporal factor.   19. The method of claim 18, wherein the degree of interest aspect is further based on new topic information determined by a date on the Internet.   The method of claim 18, wherein the degree of interest aspect is further based on information on topics considered new to the user by the advice facility.   19. The method of claim 18, wherein the degree of interest aspect is further based on information on a topic having a new aspect relative to an existing topic.   The method of claim 18, wherein the degree of interest aspect is further determined in relation to a ratio of items rated by the user to items stored by the user.   19. The method of claim 18, wherein the interest aspect is further determined by an acceleration of social activity associated with the topic information.   The method of claim 1, wherein the geographic aspect is a geographic location associated with the topic information.   The method of claim 25, wherein the geographic location is an event location.   The method of claim 25, wherein the geographic location is a store location.   26. The method of claim 25, wherein the geographical location is a restaurant location.   26. The method of claim 25, wherein the geographic location is a location of interest.   The method of claim 25, wherein the geographic location is a location of at least one product.   The method of claim 1, wherein the derived user preference is based on a rating.   The method of claim 31, wherein the rating is provided by the user.   32. The method of claim 31, wherein the rating is provided by the user's friends.   32. The method of claim 31, wherein the rating is provided by people who have similar preferences as the user.   The method of claim 31, wherein the rating is provided by an authoritative source.   The method of claim 31, wherein the rating is provided by a celebrity.   The method of claim 31, wherein the rating is inferred from user behavior.   38. The method of claim 37, wherein the user behavior is an online behavior comprising at least one of a shopping behavior, a browsing behavior, a social networking behavior, or a location-based behavior.   32. The method of claim 31, wherein the derived user preference is based on machine learning about the user's online behavior.   The method of claim 1, wherein the recommendation is at least one of an item to buy, a place to visit, an event to attend, and a place to eat.   The recommendation is based on one of a rating and a recommendation of at least one other user with a preference similar to the user from the user's current geographic location. The method described in 1.   The method of claim 41, wherein the similar preference is determined by machine learning using at least one of the other user's rating and the other user's online behavior.   The method of claim 1, wherein the recommendation is provided as part of a local discovery recommendation supply.   The method of claim 1, wherein the recommendation is stored in a list that is at least one of a wish list, a to-do list, an event list, and a transaction list.   The stored recommendation is presented to the user by the local discovery application when an aspect of the geographic location of the stored recommendation matches the current location of the user. Item 2. The method according to Item 1.   A recommendation is sent to the user based on a recommendation from at least one other user, and the advice facility determines an applicable radius around the current location of the user in using the recommendation of the other user. The method according to claim 1.   The method of claim 1, wherein the application is a local discovery application.   48. The method of claim 47, wherein the local discovery application associates at least one of news and stored recommendations with weather.   48. The method of claim 47, wherein the local discovery application associates at least one of news and a stored recommendation with a time of day.   48. The method of claim 47, wherein the local discovery application associates at least one of news and a stored recommendation with the user's social activity.   The local discovery application, wherein the advice facility displays an image based on how interested the user thinks at least one of each recommendation and each topic. 48. The method according to 47.   The displayed image is a random from left to right, top to bottom arrangement based on how much the advice facility considers the user to be interested in at least one of each recommendation and each topic. 52. The method of claim 51, wherein the method is displayed in a grid.